Created
March 9, 2022 18:43
-
-
Save ZhouXing19/1c4434754b9acd9acbcc07507a78143a to your computer and use it in GitHub Desktop.
This file has been truncated, but you can view the full file.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| diff --git a/.bazelrc b/.bazelrc | |
| index dc0596733f..2e0f49554a 100644 | |
| --- a/.bazelrc | |
| +++ b/.bazelrc | |
| @@ -26,27 +26,18 @@ build:cross --stamp | |
| build:cross --define cockroach_cross=y | |
| # Cross-compilation configurations. Add e.g. --config=crosslinux to turn these on. | |
| -# Generally these should be used for development builds. Each cross config has | |
| -# a corresponding `base` config that is the same thing but without the | |
| -# `--workspace_status_command`; if using these `base` configs, you need to | |
| -# specify an appropriate `--workspace_status_command`. These `base` configs are | |
| -# used by the release process which needs to have more control over stamping. | |
| +build:crosslinux --platforms=//build/toolchains:cross_linux | |
| build:crosslinux '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu' | |
| -build:crosslinux --config=crosslinuxbase | |
| -build:crosslinuxbase --platforms=//build/toolchains:cross_linux | |
| -build:crosslinuxbase --config=cross | |
| +build:crosslinux --config=cross | |
| +build:crosswindows --platforms=//build/toolchains:cross_windows | |
| build:crosswindows '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32' | |
| -build:crosswindows --config=crosswindowsbase | |
| -build:crosswindowsbase --platforms=//build/toolchains:cross_windows | |
| -build:crosswindowsbase --config=cross | |
| +build:crosswindows --config=cross | |
| +build:crossmacos --platforms=//build/toolchains:cross_macos | |
| build:crossmacos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19' | |
| -build:crossmacos --config=crossmacosbase | |
| -build:crossmacosbase --platforms=//build/toolchains:cross_macos | |
| -build:crossmacosbase --config=cross | |
| +build:crossmacos --config=cross | |
| +build:crosslinuxarm --platforms=//build/toolchains:cross_linux_arm | |
| build:crosslinuxarm '--workspace_status_command=./build/bazelutil/stamp.sh aarch64-unknown-linux-gnu' | |
| -build:crosslinuxarm --config=crosslinuxarmbase | |
| -build:crosslinuxarmbase --platforms=//build/toolchains:cross_linux_arm | |
| -build:crosslinuxarmbase --config=cross | |
| +build:crosslinuxarm --config=cross | |
| # Developer configurations. Add e.g. --config=devdarwinx86_64 to turn these on. | |
| # NB: This is consumed in `BUILD` files (see build/toolchains/BUILD.bazel). | |
| diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS | |
| index dada0a3861..54de5e589c 100644 | |
| --- a/.github/CODEOWNERS | |
| +++ b/.github/CODEOWNERS | |
| @@ -38,11 +38,11 @@ | |
| /pkg/sql/show_create*.go @cockroachdb/sql-syntax-prs | |
| /pkg/sql/types/ @cockroachdb/sql-syntax-prs | |
| -/pkg/sql/crdb_internal.go @cockroachdb/sql-experience | |
| -/pkg/sql/pg_catalog.go @cockroachdb/sql-experience | |
| -/pkg/sql/pgwire/ @cockroachdb/sql-experience | |
| -/pkg/sql/sem/builtins/ @cockroachdb/sql-experience | |
| -/pkg/sql/vtable/ @cockroachdb/sql-experience | |
| +/pkg/sql/crdb_internal.go @cockroachdb/sql-api-prs | |
| +/pkg/sql/pg_catalog.go @cockroachdb/sql-api-prs | |
| +/pkg/sql/pgwire/ @cockroachdb/sql-api-prs | |
| +/pkg/sql/sem/builtins/ @cockroachdb/sql-api-prs | |
| +/pkg/sql/vtable/ @cockroachdb/sql-api-prs | |
| /pkg/sql/sessiondata/ @cockroachdb/sql-experience | |
| /pkg/sql/tests/rsg_test.go @cockroachdb/sql-experience | |
| @@ -220,14 +220,12 @@ | |
| /pkg/internal/team/ @cockroachdb/test-eng | |
| /pkg/jobs/ @cockroachdb/cdc-prs | |
| /pkg/keys/ @cockroachdb/kv-prs | |
| -/pkg/keysbase/ @cockroachdb/kv-prs | |
| # Don't ping KV on updates to reserved descriptor IDs and such. | |
| /pkg/keys/constants.go @cockroachdb/kv-prs-noreview | |
| /pkg/migration/ @cockroachdb/kv-prs-noreview @cockroachdb/sql-schema | |
| /pkg/multitenant @cockroachdb/unowned | |
| /pkg/release/ @cockroachdb/dev-inf | |
| /pkg/roachpb/.gitattributes @cockroachdb/dev-inf | |
| -/pkg/roachpb/ambiguous_* @cockroachdb/kv-prs | |
| /pkg/roachpb/api* @cockroachdb/kv-prs | |
| /pkg/roachpb/batch* @cockroachdb/kv-prs | |
| /pkg/roachpb/BUILD.bazel @cockroachdb/kv-prs-noreview | |
| @@ -249,9 +247,8 @@ | |
| /pkg/roachpb/span* @cockroachdb/kv-prs | |
| /pkg/roachpb/string_test.go @cockroachdb/kv-prs | |
| /pkg/roachpb/tenant* @cockroachdb/kv-prs | |
| -/pkg/roachpb/testdata/ambi* @cockroachdb/kv-prs | |
| /pkg/roachpb/testdata/repl* @cockroachdb/kv-prs | |
| -/pkg/roachpb/version* @cockroachdb/unowned | |
| +/pkg/roachpb/version* @cockroachdb/server | |
| /pkg/roachprod/ @cockroachdb/dev-inf | |
| /pkg/rpc/ @cockroachdb/server-prs | |
| /pkg/scheduledjobs/ @cockroachdb/bulk-prs | |
| diff --git a/.github/bors.toml b/.github/bors.toml | |
| index a3aa263a50..d6a6da5a4b 100644 | |
| --- a/.github/bors.toml | |
| +++ b/.github/bors.toml | |
| @@ -8,7 +8,7 @@ status = ["GitHub CI (Cockroach)"] | |
| # r+-ed. If it's still in progress (for e.g. if CI is still running), bors will | |
| # construct the merge commit in parallel and simply wait for success right | |
| # before merging. | |
| -pr_status = ["license/cla", "blathers/release-justification-check"] | |
| +pr_status = ["license/cla"] | |
| # List of PR labels that may not be attached to a PR when it is r+-ed. | |
| block_labels = ["do-not-merge"] | |
| diff --git a/DEPS.bzl b/DEPS.bzl | |
| index ddf4b85115..631bf477bb 100644 | |
| --- a/DEPS.bzl | |
| +++ b/DEPS.bzl | |
| @@ -244,20 +244,10 @@ def go_deps(): | |
| name = "com_github_apache_thrift", | |
| build_file_proto_mode = "disable_global", | |
| importpath = "github.com/apache/thrift", | |
| - sha256 = "50d5c610df30fa2a6039394d5142382b7d9938870dfb12ef46bddfa3da250893", | |
| - strip_prefix = "github.com/apache/[email protected]", | |
| + sha256 = "f9e5418fda5dff9f5e1a892a127472fc621d417b3ee1351e53141509233fb1d5", | |
| + strip_prefix = "github.com/apache/[email protected]", | |
| urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/apache/thrift/com_github_apache_thrift-v0.16.0.zip", | |
| - ], | |
| - ) | |
| - go_repository( | |
| - name = "com_github_araddon_dateparse", | |
| - build_file_proto_mode = "disable_global", | |
| - importpath = "github.com/araddon/dateparse", | |
| - sha256 = "8fe74d5b36a9aa66c1350657009296469f8aa5b6925acd178044cd248023fc5a", | |
| - strip_prefix = "github.com/araddon/[email protected]", | |
| - urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/araddon/dateparse/com_github_araddon_dateparse-v0.0.0-20210429162001-6b43995a97de.zip", | |
| + "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/apache/thrift/com_github_apache_thrift-v0.15.0.zip", | |
| ], | |
| ) | |
| go_repository( | |
| @@ -1297,10 +1287,10 @@ def go_deps(): | |
| patches = [ | |
| "@cockroach//build/patches:com_github_cockroachdb_pebble.patch", | |
| ], | |
| - sha256 = "cc3201e4197273c3ddc0adf72ab1f800d7b09e2b9d50422cab619a854d8e4e80", | |
| - strip_prefix = "github.com/cockroachdb/[email protected]", | |
| + sha256 = "71da6a69951ab9767aa51efd34b2a4040ab655f67a5b0be87578af5a85132d26", | |
| + strip_prefix = "github.com/cockroachdb/[email protected]", | |
| urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/cockroachdb/pebble/com_github_cockroachdb_pebble-v0.0.0-20220307192532-e2b7bb844759.zip", | |
| + "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/cockroachdb/pebble/com_github_cockroachdb_pebble-v0.0.0-20220227235451-40d39da505a5.zip", | |
| ], | |
| ) | |
| go_repository( | |
| @@ -2458,10 +2448,10 @@ def go_deps(): | |
| name = "com_github_fraugster_parquet_go", | |
| build_file_proto_mode = "disable_global", | |
| importpath = "github.com/fraugster/parquet-go", | |
| - sha256 = "a9d995d60a99dac8709c614f3d2795ef4d8d79c82991b40fa703006078babf75", | |
| - strip_prefix = "github.com/fraugster/[email protected]", | |
| + sha256 = "66beb8f2218c31ca4ca3d3dbcc91c256f3379750ade924016f4179982446edd7", | |
| + strip_prefix = "github.com/fraugster/[email protected]", | |
| urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/fraugster/parquet-go/com_github_fraugster_parquet_go-v0.10.0.zip", | |
| + "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/fraugster/parquet-go/com_github_fraugster_parquet_go-v0.6.1.zip", | |
| ], | |
| ) | |
| go_repository( | |
| @@ -5329,10 +5319,10 @@ def go_deps(): | |
| name = "com_github_mattn_go_runewidth", | |
| build_file_proto_mode = "disable_global", | |
| importpath = "github.com/mattn/go-runewidth", | |
| - sha256 = "030dcf3ea6cb39b34269f119ee5fe5e9d5834daf409e1f3f48c065a45326f4a6", | |
| - strip_prefix = "github.com/mattn/[email protected]", | |
| + sha256 = "dab6e7984b913f61d02ac53f5c1875c029e591f3b4d3262dea8e95d8c28bb657", | |
| + strip_prefix = "github.com/mattn/[email protected]", | |
| urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/mattn/go-runewidth/com_github_mattn_go_runewidth-v0.0.10.zip", | |
| + "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/mattn/go-runewidth/com_github_mattn_go_runewidth-v0.0.7.zip", | |
| ], | |
| ) | |
| go_repository( | |
| @@ -6743,16 +6733,6 @@ def go_deps(): | |
| "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/retailnext/hllpp/com_github_retailnext_hllpp-v1.0.1-0.20180308014038-101a6d2f8b52.zip", | |
| ], | |
| ) | |
| - go_repository( | |
| - name = "com_github_rivo_uniseg", | |
| - build_file_proto_mode = "disable_global", | |
| - importpath = "github.com/rivo/uniseg", | |
| - sha256 = "cb701df81f36acfbb2627a78662fdcaa150ee1ac00d0796a7f3eafbdb6218128", | |
| - strip_prefix = "github.com/rivo/[email protected]", | |
| - urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/rivo/uniseg/com_github_rivo_uniseg-v0.1.0.zip", | |
| - ], | |
| - ) | |
| go_repository( | |
| name = "com_github_robertkrimen_godocdown", | |
| build_file_proto_mode = "disable_global", | |
| @@ -6943,16 +6923,6 @@ def go_deps(): | |
| "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/sclevine/agouti/com_github_sclevine_agouti-v3.0.0+incompatible.zip", | |
| ], | |
| ) | |
| - go_repository( | |
| - name = "com_github_scylladb_termtables", | |
| - build_file_proto_mode = "disable_global", | |
| - importpath = "github.com/scylladb/termtables", | |
| - sha256 = "0afd3a75417e6b708c62329cea7f2822162c8bf6ada0de094cdc032461b8f196", | |
| - strip_prefix = "github.com/scylladb/[email protected]", | |
| - urls = [ | |
| - "https://storage.googleapis.com/cockroach-godeps/gomod/github.com/scylladb/termtables/com_github_scylladb_termtables-v0.0.0-20191203121021-c4c0b6d42ff4.zip", | |
| - ], | |
| - ) | |
| go_repository( | |
| name = "com_github_sean__seed", | |
| build_file_proto_mode = "disable_global", | |
| diff --git a/Makefile b/Makefile | |
| index bf2dc0ae8d..95fce1608a 100644 | |
| --- a/Makefile | |
| +++ b/Makefile | |
| @@ -1721,6 +1721,7 @@ bins = \ | |
| bin/cockroach-oss \ | |
| bin/cockroach-short \ | |
| bin/cockroach-sql \ | |
| + bin/compile-builds \ | |
| bin/docgen \ | |
| bin/execgen \ | |
| bin/fuzz \ | |
| diff --git a/WORKSPACE b/WORKSPACE | |
| index e71ed27194..74a3198740 100644 | |
| --- a/WORKSPACE | |
| +++ b/WORKSPACE | |
| @@ -137,35 +137,16 @@ http_archive( | |
| load( | |
| "@io_bazel_rules_go//go:deps.bzl", | |
| "go_download_sdk", | |
| - "go_host_sdk", | |
| - "go_local_sdk", | |
| "go_register_toolchains", | |
| "go_rules_dependencies", | |
| ) | |
| -# To point to a mirrored artifact, use: | |
| -# | |
| go_download_sdk( | |
| name = "go_sdk", | |
| urls = ["https://storage.googleapis.com/public-bazel-artifacts/go/{}"], | |
| version = "1.17.6", | |
| ) | |
| -# To point to a local SDK path, use the following instead. We'll call the | |
| -# directory into which you cloned the Go repository $GODIR[1]. You'll have to | |
| -# first run ./make.bash from $GODIR/src to pick up any custom changes. | |
| -# | |
| -# [1]: https://go.dev/doc/contribute#testing | |
| -# | |
| -# go_local_sdk( | |
| -# name = "go_sdk", | |
| -# path = "<path to $GODIR>", | |
| -# ) | |
| - | |
| -# To use your whatever your local SDK is, use the following instead: | |
| -# | |
| -# go_host_sdk(name = "go_sdk") | |
| - | |
| go_rules_dependencies() | |
| go_register_toolchains(nogo = "@cockroach//:crdb_nogo") | |
| diff --git a/build/bazelutil/check.sh b/build/bazelutil/check.sh | |
| index 2cbe67821b..3c6d31cbfe 100755 | |
| --- a/build/bazelutil/check.sh | |
| +++ b/build/bazelutil/check.sh | |
| @@ -11,7 +11,7 @@ pkg/roachprod/vm/aws/config.go://go:generate gofmt -s -w embedded.go | |
| pkg/roachprod/vm/aws/config.go://go:generate goimports -w embedded.go | |
| pkg/roachprod/vm/aws/config.go://go:generate terraformgen -o terraform/main.tf | |
| pkg/cmd/roachtest/prometheus/prometheus.go://go:generate mockgen -package=prometheus -destination=mocks_generated_test.go . Cluster | |
| -pkg/cmd/roachtest/tests/drt.go://go:generate mockgen -package tests -destination drt_generated_test.go github.com/cockroachdb/cockroach/pkg/cmd/roachtest/prometheus Client | |
| +pkg/cmd/roachtest/tests/drt.go://go:generate mockgen -package tests -destination drt_generated_test.go . PromClient | |
| pkg/kv/kvclient/kvcoord/transport.go://go:generate mockgen -package=kvcoord -destination=mocks_generated_test.go . Transport | |
| pkg/kv/kvclient/rangecache/range_cache.go://go:generate mockgen -package=rangecachemock -destination=rangecachemock/mocks_generated.go . RangeDescriptorDB | |
| pkg/kv/kvclient/rangefeed/rangefeed.go://go:generate mockgen -destination=mocks_generated_test.go --package=rangefeed . DB | |
| diff --git a/build/bazelutil/stamp.sh b/build/bazelutil/stamp.sh | |
| index a76b504fcb..55bbd4b492 100755 | |
| --- a/build/bazelutil/stamp.sh | |
| +++ b/build/bazelutil/stamp.sh | |
| @@ -3,15 +3,6 @@ | |
| # This command is used by bazel as the workspace_status_command | |
| # to implement build stamping with git information. | |
| -# Usage: stamp.sh [target-triple] [build-channel] [build-tag] [build-type] | |
| -# All arguments are optional and have appropriate defaults. In this way, | |
| -# stamp.sh with no arguments is appropriate as the `workplace_status_command` | |
| -# for a development build. | |
| -# target-triple: defaults to the value of `cc -dumpmachine` | |
| -# build-channel: defaults to `unknown`, but can be `official-binary` | |
| -# build-tag: defaults to a value that is gleaned from `git rev-parse` | |
| -# build-type: defaults to `development`, but can be `release` | |
| - | |
| set -euo pipefail | |
| # Do not use plumbing commands, like git diff-index, in this target. Our build | |
| @@ -23,52 +14,21 @@ set -euo pipefail | |
| # For details, see the "Possible timestamp problems with diff-files?" thread on | |
| # the Git mailing list (http://marc.info/?l=git&m=131687596307197). | |
| -# Handle target-triple. | |
| +GIT_BUILD_TYPE="development" | |
| +GIT_COMMIT=$(git rev-parse HEAD) | |
| +GIT_TAG=$(git describe --tags --dirty --match=v[0-9]* 2> /dev/null || git rev-parse --short HEAD;) | |
| +GIT_UTCTIME=$(date -u '+%Y/%m/%d %H:%M:%S') | |
| + | |
| if [ -z "${1+x}" ] | |
| then | |
| TARGET_TRIPLE=$(cc -dumpmachine) | |
| else | |
| TARGET_TRIPLE="$1" | |
| - shift 1 | |
| -fi | |
| - | |
| -# Handle build-channel. | |
| -if [ -z "${1+x}" ] | |
| -then | |
| - BUILD_CHANNEL="unknown" | |
| -else | |
| - BUILD_CHANNEL="$1" | |
| - shift 1 | |
| -fi | |
| - | |
| -# Handle build-tag. | |
| -if [ -z "${1+x}" ] | |
| -then | |
| - BUILD_TAG=$(git describe --tags --dirty --match=v[0-9]* 2> /dev/null || git rev-parse --short HEAD;) | |
| -else | |
| - BUILD_TAG="$1" | |
| - shift 1 | |
| -fi | |
| - | |
| -# Handle build-type. | |
| -if [ -z "${1+x}" ] | |
| -then | |
| - BUILD_TYPE="development" | |
| -else | |
| - BUILD_TYPE="$1" | |
| - shift 1 | |
| fi | |
| -if [ "$BUILD_TYPE" = "release" ] | |
| -then | |
| - CRASH_REPORT_ENV="$BUILD_TAG" | |
| -else | |
| - CRASH_REPORT_ENV="development" | |
| -fi | |
| - | |
| -BUILD_REV=$(git rev-parse HEAD) | |
| -BUILD_UTCTIME=$(date -u '+%Y/%m/%d %H:%M:%S') | |
| - | |
| +# TODO(ricky): Also provide a way to stamp the following variables: | |
| +# - github.com/cockroachdb/cockroach/pkg/build.channel | |
| +# - github.com/cockroachdb/cockroach/pkg/util/log/logcrash.crashReportEnv | |
| # Variables beginning with "STABLE" will be written to stable-status.txt, and | |
| # others will be written to volatile-status.txt. | |
| @@ -78,11 +38,9 @@ BUILD_UTCTIME=$(date -u '+%Y/%m/%d %H:%M:%S') | |
| # * https://docs.bazel.build/versions/main/user-manual.html#workspace_status | |
| # * https://github.com/bazelbuild/rules_go/blob/master/go/core.rst#defines-and-stamping | |
| cat <<EOF | |
| -STABLE_BUILD_CHANNEL ${BUILD_CHANNEL-} | |
| +STABLE_BUILD_GIT_BUILD_TYPE ${GIT_BUILD_TYPE-} | |
| STABLE_BUILD_TARGET_TRIPLE ${TARGET_TRIPLE-} | |
| -STABLE_BUILD_TYPE ${BUILD_TYPE-} | |
| -STABLE_CRASH_REPORT_ENV ${CRASH_REPORT_ENV-} | |
| -BUILD_REV ${BUILD_REV-} | |
| -BUILD_TAG ${BUILD_TAG-} | |
| -BUILD_UTCTIME ${BUILD_UTCTIME-} | |
| +BUILD_GIT_COMMIT ${GIT_COMMIT-} | |
| +BUILD_GIT_TAG ${GIT_TAG-} | |
| +BUILD_GIT_UTCTIME ${GIT_UTCTIME-} | |
| EOF | |
| diff --git a/build/teamcity/internal/release/process/make-and-publish-build.sh b/build/release/teamcity-make-and-publish-build.sh | |
| similarity index 81% | |
| rename from build/teamcity/internal/release/process/make-and-publish-build.sh | |
| rename to build/release/teamcity-make-and-publish-build.sh | |
| index 6e74e45c1d..f4d24f5bea 100755 | |
| --- a/build/teamcity/internal/release/process/make-and-publish-build.sh | |
| +++ b/build/release/teamcity-make-and-publish-build.sh | |
| @@ -2,13 +2,13 @@ | |
| set -euo pipefail | |
| -dir="$(dirname $(dirname $(dirname $(dirname $(dirname "${0}")))))" | |
| -source "$dir/release/teamcity-support.sh" | |
| -source "$dir/teamcity-bazel-support.sh" # for run_bazel | |
| +source "$(dirname "${0}")/teamcity-support.sh" | |
| tc_start_block "Variable Setup" | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| -build_name=$(git describe --tags --dirty --match=v[0-9]* 2> /dev/null || git rev-parse --short HEAD;) | |
| +build/builder.sh make .buildinfo/tag | |
| +build_name="${TAG_NAME:-$(cat .buildinfo/tag)}" | |
| # On no match, `grep -Eo` returns 1. `|| echo""` makes the script not error. | |
| release_branch="$(echo "$build_name" | grep -Eo "^v[0-9]+\.[0-9]+" || echo"")" | |
| @@ -44,14 +44,21 @@ tc_start_block "Tag the release" | |
| git tag "${build_name}" | |
| tc_end_block "Tag the release" | |
| + | |
| +tc_start_block "Compile publish-provisional-artifacts" | |
| +build/builder.sh go install ./pkg/cmd/publish-provisional-artifacts | |
| +tc_end_block "Compile publish-provisional-artifacts" | |
| + | |
| + | |
| tc_start_block "Compile and publish S3 artifacts" | |
| -BAZEL_SUPPORT_EXTRA_DOCKER_ARGS="-e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e TC_BUILD_BRANCH=$build_name -e bucket=$bucket" run_bazel << 'EOF' | |
| -bazel build --config ci //pkg/cmd/publish-provisional-artifacts | |
| -BAZEL_BIN=$(bazel info bazel-bin --config ci) | |
| -$BAZEL_BIN/pkg/cmd/publish-provisional-artifacts/publish-provisional-artifacts_/publish-provisional-artifacts -provisional -release -bucket "$bucket" | |
| -EOF | |
| +build/builder.sh env \ | |
| + AWS_ACCESS_KEY_ID="$AWS_ACCESS_KEY_ID" \ | |
| + AWS_SECRET_ACCESS_KEY="$AWS_SECRET_ACCESS_KEY" \ | |
| + TC_BUILD_BRANCH="$build_name" \ | |
| + publish-provisional-artifacts -provisional -release -bucket "$bucket" | |
| tc_end_block "Compile and publish S3 artifacts" | |
| + | |
| tc_start_block "Make and push docker image" | |
| configure_docker_creds | |
| docker_login_with_google | |
| @@ -67,6 +74,7 @@ docker build --no-cache --tag="${gcr_repository}:${build_name}" build/deploy | |
| docker push "${gcr_repository}:${build_name}" | |
| tc_end_block "Make and push docker image" | |
| + | |
| tc_start_block "Push release tag to github.com/cockroachdb/cockroach" | |
| github_ssh_key="${GITHUB_COCKROACH_TEAMCITY_PRIVATE_SSH_KEY}" | |
| configure_git_ssh_key | |
| diff --git a/build/teamcity/internal/release/process/publish-cockroach-release.sh b/build/release/teamcity-publish-release.sh | |
| similarity index 87% | |
| rename from build/teamcity/internal/release/process/publish-cockroach-release.sh | |
| rename to build/release/teamcity-publish-release.sh | |
| index 9df13c006c..d0f7754100 100755 | |
| --- a/build/teamcity/internal/release/process/publish-cockroach-release.sh | |
| +++ b/build/release/teamcity-publish-release.sh | |
| @@ -2,11 +2,10 @@ | |
| set -euxo pipefail | |
| -dir="$(dirname $(dirname $(dirname $(dirname $(dirname "${0}")))))" | |
| -source "$dir/release/teamcity-support.sh" | |
| -source "$dir/teamcity-bazel-support.sh" # for run_bazel | |
| +source "$(dirname "${0}")/teamcity-support.sh" | |
| tc_start_block "Variable Setup" | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| # Matching the version name regex from within the cockroach code except | |
| # for the `metadata` part at the end because Docker tags don't support | |
| @@ -74,14 +73,19 @@ git tag "${build_name}" | |
| tc_end_block "Tag the release" | |
| +tc_start_block "Compile publish-provisional-artifacts" | |
| +build/builder.sh go install ./pkg/cmd/publish-provisional-artifacts | |
| +tc_end_block "Compile publish-provisional-artifacts" | |
| + | |
| + | |
| tc_start_block "Make and publish release S3 artifacts" | |
| # Using publish-provisional-artifacts here is funky. We're directly publishing | |
| # the official binaries, not provisional ones. Legacy naming. To clean up... | |
| -BAZEL_SUPPORT_EXTRA_DOCKER_ARGS="-e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e TC_BUILD_BRANCH=$build_name -e bucket=$bucket" run_bazel << 'EOF' | |
| -bazel build --config ci //pkg/cmd/publish-provisional-artifacts | |
| -BAZEL_BIN=$(bazel info bazel-bin --config ci) | |
| -$BAZEL_BIN/pkg/cmd/publish-provisional-artifacts/publish-provisional-artifacts_/publish-provisional-artifacts -provisional -release -bucket "$bucket" | |
| -EOF | |
| +build/builder.sh env \ | |
| + AWS_ACCESS_KEY_ID="$AWS_ACCESS_KEY_ID" \ | |
| + AWS_SECRET_ACCESS_KEY="$AWS_SECRET_ACCESS_KEY" \ | |
| + TC_BUILD_BRANCH="$build_name" \ | |
| + publish-provisional-artifacts -provisional -release -bucket "$bucket" | |
| tc_end_block "Make and publish release S3 artifacts" | |
| @@ -127,12 +131,11 @@ tc_start_block "Publish S3 binaries and archive as latest" | |
| # Only push the "latest" for our most recent release branch. | |
| # https://github.com/cockroachdb/cockroach/issues/41067 | |
| if [[ -n "${PUBLISH_LATEST}" && -z "${PRE_RELEASE}" ]]; then | |
| - BAZEL_SUPPORT_EXTRA_DOCKER_ARGS="-e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e TC_BUILD_BRANCH=$build_name -e bucket=$bucket" run_bazel << 'EOF' | |
| -bazel build --config ci //pkg/cmd/publish-provisional-artifacts | |
| -BAZEL_BIN=$(bazel info bazel-bin --config ci) | |
| -$BAZEL_BIN/pkg/cmd/publish-provisional-artifacts/publish-provisional-artifacts_/publish-provisional-artifacts -bless -release -bucket "$bucket" | |
| -EOF | |
| - | |
| + build/builder.sh env \ | |
| + AWS_ACCESS_KEY_ID="$AWS_ACCESS_KEY_ID" \ | |
| + AWS_SECRET_ACCESS_KEY="$AWS_SECRET_ACCESS_KEY" \ | |
| + TC_BUILD_BRANCH="$build_name" \ | |
| + publish-provisional-artifacts -bless -release -bucket "${bucket}" | |
| else | |
| echo "The latest S3 binaries and archive were _not_ updated." | |
| fi | |
| diff --git a/build/release/teamcity-support.sh b/build/release/teamcity-support.sh | |
| index aeb3cb4b8d..0518b15adf 100644 | |
| --- a/build/release/teamcity-support.sh | |
| +++ b/build/release/teamcity-support.sh | |
| @@ -1,7 +1,7 @@ | |
| # Common helpers for teamcity-*.sh scripts. | |
| # root is the absolute path to the root directory of the repository. | |
| -root="$(dirname $(dirname $(cd "$( dirname "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )))" | |
| +root=$(cd "$(dirname "$0")/../.." && pwd) | |
| source "$root/build/teamcity-common-support.sh" | |
| remove_files_on_exit() { | |
| diff --git a/build/teamcity-bless-provisional-binaries.sh b/build/teamcity-bless-provisional-binaries.sh | |
| new file mode 100755 | |
| index 0000000000..5a463ddcb5 | |
| --- /dev/null | |
| +++ b/build/teamcity-bless-provisional-binaries.sh | |
| @@ -0,0 +1,15 @@ | |
| +#!/usr/bin/env bash | |
| + | |
| +# Any arguments to this script are passed through unmodified to | |
| +# ./pkg/cmd/publish-provisional-artifacts. | |
| + | |
| +set -euxo pipefail | |
| + | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| + | |
| +build/builder.sh go install ./pkg/cmd/publish-provisional-artifacts | |
| +build/builder.sh env \ | |
| + AWS_ACCESS_KEY_ID="$AWS_ACCESS_KEY_ID" \ | |
| + AWS_SECRET_ACCESS_KEY="$AWS_SECRET_ACCESS_KEY" \ | |
| + TC_BUILD_BRANCH="$TC_BUILD_BRANCH" \ | |
| + publish-provisional-artifacts -bless "$@" | |
| diff --git a/build/teamcity-check.sh b/build/teamcity-check.sh | |
| index 1031a2c486..66ad94333f 100755 | |
| --- a/build/teamcity-check.sh | |
| +++ b/build/teamcity-check.sh | |
| @@ -6,6 +6,8 @@ source "$(dirname "${0}")/teamcity-support.sh" | |
| tc_prepare | |
| +maybe_require_release_justification | |
| + | |
| tc_start_block "Lint" | |
| # Disable ccache so that Go doesn't try to install dependencies into GOROOT, | |
| # where it doesn't have write permissions. (Using ccache busts the Go package | |
| diff --git a/build/teamcity-compile-build.sh b/build/teamcity-compile-build.sh | |
| new file mode 100755 | |
| index 0000000000..e4945b0f40 | |
| --- /dev/null | |
| +++ b/build/teamcity-compile-build.sh | |
| @@ -0,0 +1,9 @@ | |
| +#!/usr/bin/env bash | |
| + | |
| +set -euxo pipefail | |
| + | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| + | |
| +build/builder.sh go install ./pkg/cmd/compile-build | |
| +build/builder.sh env \ | |
| + compile-build | |
| diff --git a/build/teamcity-compile-builds.sh b/build/teamcity-compile-builds.sh | |
| index 15d46678f2..d1cbaf808d 100755 | |
| --- a/build/teamcity-compile-builds.sh | |
| +++ b/build/teamcity-compile-builds.sh | |
| @@ -4,7 +4,7 @@ set -euxo pipefail | |
| export BUILDER_HIDE_GOPATH_SRC=1 | |
| -build/builder.sh mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 | |
| -build/builder.sh mkrelease darwin SUFFIX=.darwin-10.9-amd64 | |
| -build/builder.sh mkrelease windows SUFFIX=.windows-6.2-amd64.exe | |
| +build/builder.sh go install ./pkg/cmd/compile-build | |
| +build/builder.sh env \ | |
| + compile-build --all --buildtype=development | |
| cp cockroach.* artifacts | |
| diff --git a/build/teamcity-publish-artifacts.sh b/build/teamcity-publish-artifacts.sh | |
| new file mode 100755 | |
| index 0000000000..7a53a07fd5 | |
| --- /dev/null | |
| +++ b/build/teamcity-publish-artifacts.sh | |
| @@ -0,0 +1,11 @@ | |
| +#!/usr/bin/env bash | |
| + | |
| +# Any arguments to this script are passed through unmodified to | |
| +# ./build/teamcity-publish-s3-binaries. | |
| + | |
| +set -euxo pipefail | |
| + | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| + | |
| +build/teamcity-publish-s3-binaries.sh "$@" | |
| + | |
| diff --git a/build/teamcity-publish-s3-binaries.sh b/build/teamcity-publish-s3-binaries.sh | |
| new file mode 100755 | |
| index 0000000000..485895b847 | |
| --- /dev/null | |
| +++ b/build/teamcity-publish-s3-binaries.sh | |
| @@ -0,0 +1,15 @@ | |
| +#!/usr/bin/env bash | |
| + | |
| +# Any arguments to this script are passed through unmodified to | |
| +# ./pkg/cmd/publish-artifacts. | |
| + | |
| +set -euxo pipefail | |
| + | |
| +export BUILDER_HIDE_GOPATH_SRC=1 | |
| + | |
| +build/builder.sh go install ./pkg/cmd/publish-artifacts | |
| +build/builder.sh env \ | |
| + AWS_ACCESS_KEY_ID="$AWS_ACCESS_KEY_ID" \ | |
| + AWS_SECRET_ACCESS_KEY="$AWS_SECRET_ACCESS_KEY" \ | |
| + TC_BUILD_BRANCH="$TC_BUILD_BRANCH" \ | |
| + publish-artifacts "$@" | |
| diff --git a/build/teamcity-support.sh b/build/teamcity-support.sh | |
| old mode 100755 | |
| new mode 100644 | |
| index 70b0afa031..260ece9f8d | |
| --- a/build/teamcity-support.sh | |
| +++ b/build/teamcity-support.sh | |
| @@ -307,6 +307,23 @@ generate_ssh_key() { | |
| fi | |
| } | |
| +maybe_require_release_justification() { | |
| + # Set this to 1 to require a "release justification" note in the commit message | |
| + # or the PR description. | |
| + require_justification=1 | |
| + if [ "$require_justification" = 1 ]; then | |
| + tc_start_block "Ensure commit message contains a release justification" | |
| + # Ensure master branch commits have a release justification. | |
| + if [[ $(git log -n1 | grep -ci "Release justification: \S\+") == 0 ]]; then | |
| + echo "Build Failed. No Release justification in the commit message or in the PR description." >&2 | |
| + echo "Commits must have a Release justification of the form:" >&2 | |
| + echo "Release justification: <some description of why this commit is safe to add to the release branch.>" >&2 | |
| + exit 1 | |
| + fi | |
| + tc_end_block "Ensure commit message contains a release justification" | |
| + fi | |
| +} | |
| + | |
| # Call this function with one argument, the error message to print if the | |
| # workspace is dirty. | |
| check_workspace_clean() { | |
| diff --git a/build/teamcity/cockroach/ci/tests/lint.sh b/build/teamcity/cockroach/ci/tests/lint.sh | |
| index 131984fe80..9255e3c10d 100755 | |
| --- a/build/teamcity/cockroach/ci/tests/lint.sh | |
| +++ b/build/teamcity/cockroach/ci/tests/lint.sh | |
| @@ -7,6 +7,8 @@ dir="$(dirname $(dirname $(dirname $(dirname $(dirname "${0}")))))" | |
| source "$dir/teamcity-support.sh" # For $root | |
| source "$dir/teamcity-bazel-support.sh" # For run_bazel | |
| +maybe_require_release_justification | |
| + | |
| tc_start_block "Run lints" | |
| run_bazel build/teamcity/cockroach/ci/tests/lint_impl.sh | |
| tc_end_block "Run lints" | |
| diff --git a/build/teamcity/cockroach/post-merge/publish-bleeding-edge.sh b/build/teamcity/cockroach/post-merge/publish-bleeding-edge.sh | |
| deleted file mode 100755 | |
| index 0adeebdfb2..0000000000 | |
| --- a/build/teamcity/cockroach/post-merge/publish-bleeding-edge.sh | |
| +++ /dev/null | |
| @@ -1,16 +0,0 @@ | |
| -#!/usr/bin/env bash | |
| - | |
| -# This script is called by the build configuration | |
| -# "Cockroach > Post Merge > Publish Bleeding Edge" in TeamCity. | |
| - | |
| -set -euxo pipefail | |
| - | |
| -dir="$(dirname $(dirname $(dirname $(dirname "${0}"))))" | |
| -source "$dir/teamcity-support.sh" | |
| -source "$dir/teamcity-bazel-support.sh" | |
| - | |
| -BAZEL_SUPPORT_EXTRA_DOCKER_ARGS="-e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY -e TC_BUILD_BRANCH" run_bazel << 'EOF' | |
| -bazel build --config ci //pkg/cmd/publish-artifacts | |
| -BAZEL_BIN=$(bazel info bazel-bin --config ci) | |
| -$BAZEL_BIN/pkg/cmd/publish-artifacts/publish-artifacts_/publish-artifacts | |
| -EOF | |
| diff --git a/cloud/kubernetes/bring-your-own-certs/client.yaml b/cloud/kubernetes/bring-your-own-certs/client.yaml | |
| index a53dbd6305..378e405c7f 100644 | |
| --- a/cloud/kubernetes/bring-your-own-certs/client.yaml | |
| +++ b/cloud/kubernetes/bring-your-own-certs/client.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/bring-your-own-certs/client.yaml | |
| # This config file demonstrates how to connect to the CockroachDB StatefulSet | |
| # defined in bring-your-own-certs-statefulset.yaml that uses certificates | |
| # created outside of Kubernetes. See that file for why you may want to use it. | |
| diff --git a/cloud/kubernetes/bring-your-own-certs/cockroachdb-statefulset.yaml b/cloud/kubernetes/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| index 51f62ba81d..86c1868de1 100644 | |
| --- a/cloud/kubernetes/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| +++ b/cloud/kubernetes/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| # This config file defines a CockroachDB StatefulSet that uses certificates | |
| # created outside of Kubernetes. You may want to use it if you want to use a | |
| # different certificate authority from the one being used by Kubernetes or if | |
| diff --git a/cloud/kubernetes/client-secure.yaml b/cloud/kubernetes/client-secure.yaml | |
| index 577e8c1bfc..d0c7967dbd 100644 | |
| --- a/cloud/kubernetes/client-secure.yaml | |
| +++ b/cloud/kubernetes/client-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/client-secure.yaml | |
| apiVersion: v1 | |
| kind: Pod | |
| metadata: | |
| diff --git a/cloud/kubernetes/cluster-init-secure.yaml b/cloud/kubernetes/cluster-init-secure.yaml | |
| index 840ee1f3fc..0154e8680e 100644 | |
| --- a/cloud/kubernetes/cluster-init-secure.yaml | |
| +++ b/cloud/kubernetes/cluster-init-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/cluster-init-secure.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/cluster-init.yaml b/cloud/kubernetes/cluster-init.yaml | |
| index e9dd778dd5..d33a6fb7d0 100644 | |
| --- a/cloud/kubernetes/cluster-init.yaml | |
| +++ b/cloud/kubernetes/cluster-init.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/cluster-init.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/cockroachdb-statefulset-secure.yaml | |
| index 5056a8df5e..84e1979fd1 100644 | |
| --- a/cloud/kubernetes/cockroachdb-statefulset-secure.yaml | |
| +++ b/cloud/kubernetes/cockroachdb-statefulset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/cockroachdb-statefulset-secure.yaml | |
| apiVersion: v1 | |
| kind: ServiceAccount | |
| metadata: | |
| diff --git a/cloud/kubernetes/cockroachdb-statefulset.yaml b/cloud/kubernetes/cockroachdb-statefulset.yaml | |
| index 473617f0e7..10e1fd591e 100644 | |
| --- a/cloud/kubernetes/cockroachdb-statefulset.yaml | |
| +++ b/cloud/kubernetes/cockroachdb-statefulset.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/cockroachdb-statefulset.yaml | |
| apiVersion: v1 | |
| kind: Service | |
| metadata: | |
| diff --git a/cloud/kubernetes/multiregion/client-secure.yaml b/cloud/kubernetes/multiregion/client-secure.yaml | |
| index edf7fbfa64..cf57e92628 100644 | |
| --- a/cloud/kubernetes/multiregion/client-secure.yaml | |
| +++ b/cloud/kubernetes/multiregion/client-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/multiregion/client-secure.yaml | |
| apiVersion: v1 | |
| kind: Pod | |
| metadata: | |
| diff --git a/cloud/kubernetes/multiregion/cluster-init-secure.yaml b/cloud/kubernetes/multiregion/cluster-init-secure.yaml | |
| index 3b44f3634e..c268d3f3e9 100644 | |
| --- a/cloud/kubernetes/multiregion/cluster-init-secure.yaml | |
| +++ b/cloud/kubernetes/multiregion/cluster-init-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/multiregion/cluster-init-secure.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/multiregion/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/multiregion/cockroachdb-statefulset-secure.yaml | |
| index 3f32b0515e..891ea8ca9a 100644 | |
| --- a/cloud/kubernetes/multiregion/cockroachdb-statefulset-secure.yaml | |
| +++ b/cloud/kubernetes/multiregion/cockroachdb-statefulset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/multiregion/cockroachdb-statefulset-secure.yaml | |
| apiVersion: v1 | |
| kind: ServiceAccount | |
| metadata: | |
| diff --git a/cloud/kubernetes/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml b/cloud/kubernetes/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| index 24fea374d4..1513692774 100644 | |
| --- a/cloud/kubernetes/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| +++ b/cloud/kubernetes/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| apiVersion: v1 | |
| kind: ServiceAccount | |
| metadata: | |
| diff --git a/cloud/kubernetes/performance/cockroachdb-daemonset-insecure.yaml b/cloud/kubernetes/performance/cockroachdb-daemonset-insecure.yaml | |
| index e931f6ed8b..7e3add4cf9 100644 | |
| --- a/cloud/kubernetes/performance/cockroachdb-daemonset-insecure.yaml | |
| +++ b/cloud/kubernetes/performance/cockroachdb-daemonset-insecure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/performance/cockroachdb-daemonset-insecure.yaml | |
| # This configuration file sets up a DaemonSet running CockroachDB in insecure | |
| # mode. For more information on why you might want to use a DaemonSet instead | |
| # of a StatefulSet, see our docs: | |
| diff --git a/cloud/kubernetes/performance/cockroachdb-daemonset-secure.yaml b/cloud/kubernetes/performance/cockroachdb-daemonset-secure.yaml | |
| index 5174ccad09..4ba05cef70 100644 | |
| --- a/cloud/kubernetes/performance/cockroachdb-daemonset-secure.yaml | |
| +++ b/cloud/kubernetes/performance/cockroachdb-daemonset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/performance/cockroachdb-daemonset-secure.yaml | |
| # This configuration file sets up a secure DaemonSet running CockroachDB. | |
| # For more information on why you might want to use a DaemonSet instead | |
| # of a StatefulSet, see our docs: | |
| diff --git a/cloud/kubernetes/performance/cockroachdb-statefulset-insecure.yaml b/cloud/kubernetes/performance/cockroachdb-statefulset-insecure.yaml | |
| index 79c20879e8..50a550c505 100644 | |
| --- a/cloud/kubernetes/performance/cockroachdb-statefulset-insecure.yaml | |
| +++ b/cloud/kubernetes/performance/cockroachdb-statefulset-insecure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/performance/cockroachdb-statefulset-insecure.yaml | |
| # This configuration file sets up an insecure StatefulSet running CockroachDB with | |
| # tweaks to make it more performant than our default configuration files. All | |
| # changes from the default insecure configuration have been marked with a comment | |
| diff --git a/cloud/kubernetes/performance/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/performance/cockroachdb-statefulset-secure.yaml | |
| index 2236ba9937..f7334e3add 100644 | |
| --- a/cloud/kubernetes/performance/cockroachdb-statefulset-secure.yaml | |
| +++ b/cloud/kubernetes/performance/cockroachdb-statefulset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/performance/cockroachdb-statefulset-secure.yaml | |
| # This configuration file sets up a secure StatefulSet running CockroachDB with | |
| # tweaks to make it more performant than our default configuration files. All | |
| # changes from the default secure configuration have been marked with a comment | |
| diff --git a/cloud/kubernetes/templates/bring-your-own-certs/client.yaml b/cloud/kubernetes/templates/bring-your-own-certs/client.yaml | |
| deleted file mode 100644 | |
| index d486a5493b..0000000000 | |
| --- a/cloud/kubernetes/templates/bring-your-own-certs/client.yaml | |
| +++ /dev/null | |
| @@ -1,35 +0,0 @@ | |
| -# This config file demonstrates how to connect to the CockroachDB StatefulSet | |
| -# defined in bring-your-own-certs-statefulset.yaml that uses certificates | |
| -# created outside of Kubernetes. See that file for why you may want to use it. | |
| -# You should be able to adapt the core ideas to deploy your own custom | |
| -# applications and connect them to the database similarly. | |
| -# | |
| -# The pod that this file defines will sleep in the cluster not using any | |
| -# resources. After creating the pod, you can use it to open up a SQL shell to | |
| -# the database by running: | |
| -# | |
| -# kubectl exec -it cockroachdb-client-secure -- ./cockroach sql --url="postgres://root@cockroachdb-public:26257/?sslmode=verify-full&sslcert=/cockroach-certs/client.root.crt&sslkey=/cockroach-certs/client.root.key&sslrootcert=/cockroach-certs/ca.crt" | |
| -apiVersion: v1 | |
| -kind: Pod | |
| -metadata: | |
| - name: cockroachdb-client-secure | |
| - labels: | |
| - app: cockroachdb-client | |
| -spec: | |
| - serviceAccountName: cockroachdb | |
| - containers: | |
| - - name: cockroachdb-client | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - # Keep a pod open indefinitely so kubectl exec can be used to get a shell to it | |
| - # and run cockroach client commands, such as cockroach sql, cockroach node status, etc. | |
| - command: | |
| - - sleep | |
| - - "2147483648" # 2^31 | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - volumes: | |
| - - name: client-certs | |
| - secret: | |
| - secretName: cockroachdb.client.root | |
| - defaultMode: 256 | |
| diff --git a/cloud/kubernetes/templates/bring-your-own-certs/cockroachdb-statefulset.yaml b/cloud/kubernetes/templates/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| deleted file mode 100644 | |
| index bbfd1c264c..0000000000 | |
| --- a/cloud/kubernetes/templates/bring-your-own-certs/cockroachdb-statefulset.yaml | |
| +++ /dev/null | |
| @@ -1,244 +0,0 @@ | |
| -# This config file defines a CockroachDB StatefulSet that uses certificates | |
| -# created outside of Kubernetes. You may want to use it if you want to use a | |
| -# different certificate authority from the one being used by Kubernetes or if | |
| -# your Kubernetes cluster doesn't fully support certificate-signing requests | |
| -# (e.g. as of July 2018, EKS doesn't work properly). | |
| -# | |
| -# To use this config file, first set up your certificates and load them into | |
| -# your Kubernetes cluster as Secrets using the commands below: | |
| -# | |
| -# mkdir certs | |
| -# mkdir my-safe-directory | |
| -# cockroach cert create-ca --certs-dir=certs --ca-key=my-safe-directory/ca.key | |
| -# cockroach cert create-client root --certs-dir=certs --ca-key=my-safe-directory/ca.key | |
| -# kubectl create secret generic cockroachdb.client.root --from-file=certs | |
| -# cockroach cert create-node --certs-dir=certs --ca-key=my-safe-directory/ca.key localhost 127.0.0.1 cockroachdb-public cockroachdb-public.default cockroachdb-public.default.svc.cluster.local *.cockroachdb *.cockroachdb.default *.cockroachdb.default.svc.cluster.local | |
| -# kubectl create secret generic cockroachdb.node --from-file=certs | |
| -# kubectl create -f bring-your-own-certs-statefulset.yaml | |
| -# kubectl exec -it cockroachdb-0 -- /cockroach/cockroach init --certs-dir=/cockroach/cockroach-certs | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - - name: GOMAXPROCS | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.cpu | |
| - divisor: "1" | |
| - - name: MEMORY_LIMIT_MIB | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.memory | |
| - divisor: "1Mi" | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - exec | |
| - /cockroach/cockroach | |
| - start | |
| - --logtostderr | |
| - --certs-dir /cockroach/cockroach-certs | |
| - --advertise-host $(hostname -f) | |
| - --http-addr 0.0.0.0 | |
| - --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb | |
| - --cache $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - --max-sql-memory $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - secret: | |
| - secretName: cockroachdb.node | |
| - defaultMode: 256 | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/client-secure.yaml b/cloud/kubernetes/templates/client-secure.yaml | |
| deleted file mode 100644 | |
| index 0df772c510..0000000000 | |
| --- a/cloud/kubernetes/templates/client-secure.yaml | |
| +++ /dev/null | |
| @@ -1,48 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Pod | |
| -metadata: | |
| - name: cockroachdb-client-secure | |
| - labels: | |
| - app: cockroachdb-client | |
| -spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cockroachdb-client | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - # Keep a pod open indefinitely so kubectl exec can be used to get a shell to it | |
| - # and run cockroach client commands, such as cockroach sql, cockroach node status, etc. | |
| - command: | |
| - - sleep | |
| - - "2147483648" # 2^31 | |
| - # This pod isn't doing anything important, so don't bother waiting to terminate it. | |
| - terminationGracePeriodSeconds: 0 | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/cluster-init-secure.yaml b/cloud/kubernetes/templates/cluster-init-secure.yaml | |
| deleted file mode 100644 | |
| index 830f2928a9..0000000000 | |
| --- a/cloud/kubernetes/templates/cluster-init-secure.yaml | |
| +++ /dev/null | |
| @@ -1,49 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init-secure | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--certs-dir=/cockroach-certs" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/cluster-init.yaml b/cloud/kubernetes/templates/cluster-init.yaml | |
| deleted file mode 100644 | |
| index 07a43d8456..0000000000 | |
| --- a/cloud/kubernetes/templates/cluster-init.yaml | |
| +++ /dev/null | |
| @@ -1,19 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--insecure" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| diff --git a/cloud/kubernetes/templates/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/templates/cockroachdb-statefulset-secure.yaml | |
| deleted file mode 100644 | |
| index 86e1682cb8..0000000000 | |
| --- a/cloud/kubernetes/templates/cockroachdb-statefulset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,285 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - # Init containers are run only once in the lifetime of a pod, before | |
| - # it's started up for the first time. It has to exit successfully | |
| - # before the pod's main containers are allowed to start. | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # All addresses used to contact a node must be specified in the --addresses arg. | |
| - # | |
| - # In addition to the node certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=node -addresses=localhost,127.0.0.1,$(hostname -f),$(hostname -f|cut -f 1-2 -d '.'),cockroachdb-public,cockroachdb-public.$(hostname -f|cut -f 3- -d '.'),cockroachdb-public.$(hostname -f|cut -f 3-4 -d '.'),cockroachdb-public.$(hostname -f|cut -f 3 -d '.') -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: certs | |
| - mountPath: /cockroach-certs | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - - name: GOMAXPROCS | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.cpu | |
| - divisor: "1" | |
| - - name: MEMORY_LIMIT_MIB | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.memory | |
| - divisor: "1Mi" | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - # Memory caches are set as a fraction of the pod's memory limit. | |
| - - exec | |
| - /cockroach/cockroach | |
| - start | |
| - --logtostderr | |
| - --certs-dir /cockroach/cockroach-certs | |
| - --advertise-host $(hostname -f) | |
| - --http-addr 0.0.0.0 | |
| - --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb | |
| - --cache $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - --max-sql-memory $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - emptyDir: {} | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/cockroachdb-statefulset.yaml b/cloud/kubernetes/templates/cockroachdb-statefulset.yaml | |
| deleted file mode 100644 | |
| index f29bc741e2..0000000000 | |
| --- a/cloud/kubernetes/templates/cockroachdb-statefulset.yaml | |
| +++ /dev/null | |
| @@ -1,181 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-insecure | |
| - - name: GOMAXPROCS | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.cpu | |
| - divisor: "1" | |
| - - name: MEMORY_LIMIT_MIB | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.memory | |
| - divisor: "1Mi" | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - exec | |
| - /cockroach/cockroach | |
| - start | |
| - --logtostderr | |
| - --insecure | |
| - --advertise-host $(hostname -f) | |
| - --http-addr 0.0.0.0 | |
| - --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb | |
| - --cache $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - --max-sql-memory $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/multiregion/client-secure.yaml b/cloud/kubernetes/templates/multiregion/client-secure.yaml | |
| deleted file mode 100644 | |
| index c72a38f8fe..0000000000 | |
| --- a/cloud/kubernetes/templates/multiregion/client-secure.yaml | |
| +++ /dev/null | |
| @@ -1,27 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Pod | |
| -metadata: | |
| - name: cockroachdb-client-secure | |
| - labels: | |
| - app: cockroachdb-client | |
| -spec: | |
| - serviceAccountName: cockroachdb | |
| - containers: | |
| - - name: cockroachdb-client | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - # Keep a pod open indefinitely so kubectl exec can be used to get a shell to it | |
| - # and run cockroach client commands, such as cockroach sql, cockroach node status, etc. | |
| - command: | |
| - - sleep | |
| - - "2147483648" # 2^31 | |
| - # This pod isn't doing anything important, so don't bother waiting to terminate it. | |
| - terminationGracePeriodSeconds: 0 | |
| - volumes: | |
| - - name: client-certs | |
| - secret: | |
| - secretName: cockroachdb.client.root | |
| - defaultMode: 256 | |
| diff --git a/cloud/kubernetes/templates/multiregion/cluster-init-secure.yaml b/cloud/kubernetes/templates/multiregion/cluster-init-secure.yaml | |
| deleted file mode 100644 | |
| index 015c1c11d9..0000000000 | |
| --- a/cloud/kubernetes/templates/multiregion/cluster-init-secure.yaml | |
| +++ /dev/null | |
| @@ -1,28 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init-secure | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--certs-dir=/cockroach-certs" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| - volumes: | |
| - - name: client-certs | |
| - secret: | |
| - secretName: cockroachdb.client.root | |
| - defaultMode: 256 | |
| diff --git a/cloud/kubernetes/templates/multiregion/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/templates/multiregion/cockroachdb-statefulset-secure.yaml | |
| deleted file mode 100644 | |
| index cd6379acda..0000000000 | |
| --- a/cloud/kubernetes/templates/multiregion/cockroachdb-statefulset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,248 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-multiregion | |
| - - name: GOMAXPROCS | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.cpu | |
| - divisor: "1" | |
| - - name: MEMORY_LIMIT_MIB | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.memory | |
| - divisor: "1Mi" | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - exec | |
| - /cockroach/cockroach | |
| - start | |
| - --logtostderr | |
| - --certs-dir /cockroach/cockroach-certs | |
| - --advertise-host $(hostname -f) | |
| - --http-addr 0.0.0.0 | |
| - --join JOINLIST | |
| - --locality LOCALITYLIST | |
| - --cache $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - --max-sql-memory $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - secret: | |
| - secretName: cockroachdb.node | |
| - defaultMode: 256 | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml b/cloud/kubernetes/templates/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| deleted file mode 100644 | |
| index e829be2460..0000000000 | |
| --- a/cloud/kubernetes/templates/multiregion/eks/cockroachdb-statefulset-secure-eks.yaml | |
| +++ /dev/null | |
| @@ -1,282 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| - # TODO: Use this field to specify a namespace other than "default" in which to deploy CockroachDB (e.g., us-east-1). | |
| - # namespace: <cluster-namespace> | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - # This init container is used to determine the availability zones of the Cockroach pods. The AZs are used to define --locality when starting Cockroach nodes. | |
| - initContainers: | |
| - - command: | |
| - - sh | |
| - - -ecx | |
| - - echo "aws-$(curl http://169.254.169.254/latest/meta-data/placement/availability-zone/)" | |
| - > /etc/cockroach-env/zone | |
| - image: byrnedo/alpine-curl:0.1 | |
| - imagePullPolicy: IfNotPresent | |
| - name: locality-container | |
| - resources: {} | |
| - terminationMessagePath: /dev/termination-log | |
| - terminationMessagePolicy: File | |
| - volumeMounts: | |
| - - mountPath: /etc/cockroach-env | |
| - name: cockroach-env | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - - name: cockroach-env | |
| - mountPath: /etc/cockroach-env | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-multiregion | |
| - - name: GOMAXPROCS | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.cpu | |
| - divisor: "1" | |
| - - name: MEMORY_LIMIT_MIB | |
| - valueFrom: | |
| - resourceFieldRef: | |
| - resource: limits.memory | |
| - divisor: "1Mi" | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - exec | |
| - /cockroach/cockroach | |
| - start | |
| - --logtostderr | |
| - --certs-dir /cockroach/cockroach-certs | |
| - --advertise-host $(hostname -f) | |
| - --http-addr 0.0.0.0 | |
| - # TODO: Replace the placeholder values in --join and --locality with the namespace of the CockroachDB cluster in each region (e.g., us-east-1). | |
| - # --join cockroachdb-0.cockroachdb.<cluster-namespace-1>,cockroachdb-1.cockroachdb.<cluster-namespace-1>,cockroachdb-2.cockroachdb.<cluster-namespace-1>,cockroachdb-0.cockroachdb.<cluster-namespace-2>,cockroachdb-1.cockroachdb.<cluster-namespace-2>,cockroachdb-2.cockroachdb.<cluster-namespace-2>,cockroachdb-0.cockroachdb.<cluster-namespace-3>,cockroachdb-1.cockroachdb.<cluster-namespace-3>,cockroachdb-2.cockroachdb.<cluster-namespace-3> | |
| - # --locality=region=<cluster-namespace-1>,az=$(cat /etc/cockroach-env/zone),dns=$(hostname -f) | |
| - --cache $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - --max-sql-memory $(expr $MEMORY_LIMIT_MIB / 4)MiB | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - secret: | |
| - secretName: cockroachdb.node | |
| - defaultMode: 256 | |
| - - name: cockroach-env | |
| - emptyDir: {} | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/performance/cockroachdb-daemonset-insecure.yaml b/cloud/kubernetes/templates/performance/cockroachdb-daemonset-insecure.yaml | |
| deleted file mode 100644 | |
| index f07af4dbc8..0000000000 | |
| --- a/cloud/kubernetes/templates/performance/cockroachdb-daemonset-insecure.yaml | |
| +++ /dev/null | |
| @@ -1,142 +0,0 @@ | |
| -# This configuration file sets up a DaemonSet running CockroachDB in insecure | |
| -# mode. For more information on why you might want to use a DaemonSet instead | |
| -# of a StatefulSet, see our docs: | |
| -# https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html#running-in-a-daemonset | |
| -# | |
| -# To use this file, customize the parts labeled "TODO" before running: | |
| -# kubectl create -f cockroachdb-daemonset-insecure.yaml | |
| -# | |
| -# Initialize the cluster by picking one of the CockroachDB pod names from | |
| -# the output of `kubectl get pods`, then run: | |
| -# kubectl exec -it <pod-name> -- ./cockroach init --insecure | |
| -# | |
| -# If you're interested in using a DaemonSet in secure mode instead, please see | |
| -# cockroachdb-daemonset-secure.yaml. | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: DaemonSet | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - # TODO: Remove the nodeSelector section if you want CockroachDB to run on all nodes in your cluster. | |
| - # To give nodes this label, run: | |
| - # kubectl label node <node-name> app=cockroachdb | |
| - nodeSelector: | |
| - app: cockroachdb | |
| - # Tolerations allow CockroachDB to run on Kubernetes nodes that other pods won't be allowed on. | |
| - # To set up nodes to be dedicated to CockroachDB, you must "taint" them by running: | |
| - # kubectl taint node <node-name> app=cockroachdb:NoSchedule | |
| - # If you don't set up any such taints, these tolerations will have no effect. | |
| - tolerations: | |
| - - key: "app" | |
| - operator: "Equal" | |
| - value: "cockroachdb" | |
| - effect: "NoSchedule" | |
| - # NOTE: Running with `hostNetwork: true` means that CockroachDB will use | |
| - # the host machines' IP address and hostname, and that nothing else on | |
| - # the machines will be able to use the same ports. | |
| - hostNetwork: true | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: If you configured taints to give CockroachDB exclusive access to nodes, feel free | |
| - # to remove the requests and limits sections. If you didn't, you'll need to change these to | |
| - # appropriate values for the hardware that you're running. You can see the amount of | |
| - # allocatable resources on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - resources: | |
| - requests: | |
| - cpu: "16" | |
| - memory: "8Gi" | |
| - limits: | |
| - # NOTE: Unless you have enabled the non-default Static CPU Management Policy | |
| - # and are using an integer number of CPUs, we don't recommend setting a CPU limit. | |
| - # See: | |
| - # https://kubernetes.io/docs/tasks/administer-cluster/cpu-management-policies/#static-policy | |
| - # https://github.com/kubernetes/kubernetes/issues/51135 | |
| - #cpu: "16" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - hostPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - hostPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTP | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTP | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-insecure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # TODO: Replace "YOUR_IP_ADDR1_HERE,YOUR_IP_ADDR2_HERE,YOUR_IP_ADDR3_HERE" with a list of a few of the IP addresses of the machines on which CockroachDB will be running. | |
| - - "exec /cockroach/cockroach start --logtostderr --insecure --http-addr 0.0.0.0 --cache 25% --max-sql-memory 25% --join=YOUR_IP_ADDR1_HERE,YOUR_IP_ADDR2_HERE,YOUR_IP_ADDR3_HERE" | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - hostPath: | |
| - # TODO: Replace "YOUR_FILESYSTEM_PATH_HERE" with the path where you want CockroachDB's data stored on your Kubernetes nodes. | |
| - path: YOUR_FILESYSTEM_PATH_HERE | |
| diff --git a/cloud/kubernetes/templates/performance/cockroachdb-daemonset-secure.yaml b/cloud/kubernetes/templates/performance/cockroachdb-daemonset-secure.yaml | |
| deleted file mode 100644 | |
| index aa3e3a1eee..0000000000 | |
| --- a/cloud/kubernetes/templates/performance/cockroachdb-daemonset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,262 +0,0 @@ | |
| -# This configuration file sets up a secure DaemonSet running CockroachDB. | |
| -# For more information on why you might want to use a DaemonSet instead | |
| -# of a StatefulSet, see our docs: | |
| -# https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html#running-in-a-daemonset | |
| -# | |
| -# To use this file, customize the parts labeled "TODO" before running: | |
| -# kubectl create -f cockroachdb-daemonset-secure.yaml | |
| -# | |
| -# You will then have to approve certificate signing requests and initialize the | |
| -# cluster as described in the parent directory's README.md file. In order for | |
| -# the initialization step to work, note that you will need to change the | |
| -# address used by the cluster-init-secure.yaml file on the | |
| -# "--host=cockroachdb-0.cockroach" line from "cockroachdb-0.cockroach" to the | |
| -# address of one of your nodes. | |
| -# | |
| -# If you're interested in using a DaemonSet in insecure mode instead, please | |
| -# see cockroachdb-daemonset-insecure.yaml. | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: DaemonSet | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - # TODO: Remove the nodeSelector section if you want CockroachDB to run on all nodes in your cluster. | |
| - # To give nodes this label, run: | |
| - # kubectl label node <node-name> app=cockroachdb | |
| - nodeSelector: | |
| - app: cockroachdb | |
| - # Tolerations allow CockroachDB to run on Kubernetes nodes that other pods won't be allowed on. | |
| - # To set up nodes to be dedicated to CockroachDB, you must "taint" them by running: | |
| - # kubectl taint node <node-name> app=cockroachdb:NoSchedule | |
| - # If you don't set up any such taints, these tolerations will have no effect. | |
| - tolerations: | |
| - - key: "app" | |
| - operator: "Equal" | |
| - value: "cockroachdb" | |
| - effect: "NoSchedule" | |
| - # NOTE: Running with `hostNetwork: true` means that CockroachDB will use | |
| - # the host machines' IP address and hostname, and that nothing else on | |
| - # the machines will be able to use the same ports. | |
| - hostNetwork: true | |
| - # Init containers are run only once in the lifetime of a pod, before | |
| - # it's started up for the first time. It has to exit successfully | |
| - # before the pod's main containers are allowed to start. | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # All addresses used to contact a node must be specified in the --addresses arg. | |
| - # | |
| - # In addition to the node certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=node -addresses=localhost,127.0.0.1,$(hostname),$(hostname -f),$(hostname -i),cockroachdb-public,cockroachdb-public.${POD_NAMESPACE}.svc.cluster.local,cockroachdb-public.${POD_NAMESPACE}.svc,cockroachdb-public.${POD_NAMESPACE} -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: certs | |
| - mountPath: /cockroach-certs | |
| - # NOTE: If you are running clients that generate heavy load, you may find | |
| - # it useful to copy this anti-affinity policy into the client pods' | |
| - # configurations as well to avoid running them on the same machines as | |
| - # CockroachDB and interfering with each other's performance. | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: If you configured taints to give CockroachDB exclusive access to nodes, feel free | |
| - # to remove the requests and limits sections. If you didn't, you'll need to change these to | |
| - # appropriate values for the hardware that you're running. You can see the amount of | |
| - # allocatable resources on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - resources: | |
| - requests: | |
| - cpu: "16" | |
| - memory: "8Gi" | |
| - limits: | |
| - # NOTE: Unless you have enabled the non-default Static CPU Management Policy | |
| - # and are using an integer number of CPUs, we don't recommend setting a CPU limit. | |
| - # See: | |
| - # https://kubernetes.io/docs/tasks/administer-cluster/cpu-management-policies/#static-policy | |
| - # https://github.com/kubernetes/kubernetes/issues/51135 | |
| - #cpu: "16" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - hostPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - hostPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # TODO: Replace "YOUR_IP_ADDR1_HERE,YOUR_IP_ADDR2_HERE,YOUR_IP_ADDR3_HERE" with a list of a few of the IP addresses or hostnames of the machines on which CockroachDB will be running. | |
| - - "exec /cockroach/cockroach start --logtostderr --certs-dir /cockroach/cockroach-certs --http-addr 0.0.0.0 --cache 25% --max-sql-memory 25% --join=YOUR_IP_ADDR1_HERE,YOUR_IP_ADDR2_HERE,YOUR_IP_ADDR3_HERE" | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - hostPath: | |
| - # TODO: Replace "YOUR_FILESYSTEM_PATH_HERE" with the path where you want CockroachDB's data stored on your Kubernetes nodes. | |
| - path: YOUR_FILESYSTEM_PATH_HERE | |
| - - name: certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/performance/cockroachdb-statefulset-insecure.yaml b/cloud/kubernetes/templates/performance/cockroachdb-statefulset-insecure.yaml | |
| deleted file mode 100644 | |
| index f3cb1017cf..0000000000 | |
| --- a/cloud/kubernetes/templates/performance/cockroachdb-statefulset-insecure.yaml | |
| +++ /dev/null | |
| @@ -1,215 +0,0 @@ | |
| -# This configuration file sets up an insecure StatefulSet running CockroachDB with | |
| -# tweaks to make it more performant than our default configuration files. All | |
| -# changes from the default insecure configuration have been marked with a comment | |
| -# starting with "NOTE" or "TODO". | |
| -# | |
| -# Beware that this configuration is quite insecure. By default, it will make | |
| -# CockroachDB accessible on port 26257 on your Kubernetes nodes' network | |
| -# interfaces, meaning that if your nodes are reachable from the Internet, then | |
| -# this CockroachDB cluster will be too. To disable this behavior, remove the | |
| -# `hostNetwork` configuration field below. | |
| -# | |
| -# To use this file, customize all the parts labeled "TODO" before running: | |
| -# kubectl create -f cockroachdb-statefulset-insecure.yaml | |
| -# | |
| -# You will then have to initialize the cluster as described in the parent | |
| -# directory's README.md file. | |
| -# | |
| -# If you don't see any pods being created, it's possible that your cluster was | |
| -# not able to meet the resource requests asked for, whether it was the amount | |
| -# of CPU, memory, or disk or the disk type. To find information about why pods | |
| -# haven't been created, you can run: | |
| -# kubectl get events | |
| -# | |
| -# For more information on improving CockroachDB performance in Kubernetes, see | |
| -# our docs: | |
| -# https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - # NOTE: Running with `hostNetwork: true` means that CockroachDB will use | |
| - # the host machines' IP address and hostname, and that nothing else on | |
| - # the machines will be able to use the same ports. This means that only 1 | |
| - # CockroachDB pod will ever be schedulable on the same machine, because | |
| - # otherwise their ports would conflict. | |
| - # | |
| - # If your client pods generate a lot of network traffic to and from the | |
| - # CockroachDB cluster, you may see a benefit to doing the same thing in | |
| - # their configurations. | |
| - hostNetwork: true | |
| - dnsPolicy: ClusterFirstWithHostNet | |
| - # NOTE: If you are running clients that generate heavy load, you may find | |
| - # it useful to copy this anti-affinity policy into the client pods' | |
| - # configurations as well to avoid running them on the same machines as | |
| - # CockroachDB and interfering with each other's performance. | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - # NOTE: Always use the most recent version of CockroachDB for the best | |
| - # performance and reliability. | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-insecure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - "exec /cockroach/cockroach start --logtostderr --insecure --advertise-host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - # TODO: This specifically asks for a storage class with the name "ssd". A | |
| - # storage class of this name doesn't exist by default. See our docs for | |
| - # more information on how to create an optimized storage class for use here: | |
| - # https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html#disk-type | |
| - storageClassName: ssd | |
| - resources: | |
| - requests: | |
| - # TODO: This asks for a fairly large disk by default because on | |
| - # certain popular clouds there is a direct correlation between disk | |
| - # size and the IOPS provisioned to the disk. Change this as necessary | |
| - # to suit your needs, but be aware that smaller disks will typically | |
| - # mean worse performance. | |
| - storage: 1024Gi | |
| diff --git a/cloud/kubernetes/templates/performance/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/templates/performance/cockroachdb-statefulset-secure.yaml | |
| deleted file mode 100644 | |
| index fd9c215dcd..0000000000 | |
| --- a/cloud/kubernetes/templates/performance/cockroachdb-statefulset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,312 +0,0 @@ | |
| -# This configuration file sets up a secure StatefulSet running CockroachDB with | |
| -# tweaks to make it more performant than our default configuration files. All | |
| -# changes from the default secure configuration have been marked with a comment | |
| -# starting with "NOTE" or "TODO". | |
| -# | |
| -# To use it, customize all the parts of the file labeled "TODO" before running: | |
| -# kubectl create -f cockroachdb-statefulset-secure.yaml | |
| -# | |
| -# You will then have to approve certificate-signing requests and initialize the | |
| -# cluster as described in the parent directory's README.md file. | |
| -# | |
| -# If you don't see any pods being created, it's possible that your cluster was | |
| -# not able to meet the resource requests asked for, whether it was the amount | |
| -# of CPU, memory, or disk or the disk type. To find information about why pods | |
| -# haven't been created, you can run: | |
| -# kubectl get events | |
| -# | |
| -# For more information on improving CockroachDB performance in Kubernetes, see | |
| -# our docs: | |
| -# https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # Use this annotation in addition to the actual publishNotReadyAddresses | |
| - # field below because the annotation will stop being respected soon but the | |
| - # field is broken in some versions of Kubernetes: | |
| - # https://github.com/kubernetes/kubernetes/issues/58662 | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - # We want all pods in the StatefulSet to have their addresses published for | |
| - # the sake of the other CockroachDB pods even before they're ready, since they | |
| - # have to be able to talk to each other in order to become ready. | |
| - publishNotReadyAddresses: true | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - # NOTE: Running with `hostNetwork: true` means that CockroachDB will use | |
| - # the host machines' IP address and hostname, and that nothing else on | |
| - # the machines will be able to use the same ports. This means that only 1 | |
| - # CockroachDB pod will ever be schedulable on the same machine, because | |
| - # otherwise their ports would conflict. | |
| - # | |
| - # If your client pods generate a lot of network traffic to and from the | |
| - # CockroachDB cluster, you may see a benefit to doing the same thing in | |
| - # their configurations. | |
| - hostNetwork: true | |
| - dnsPolicy: ClusterFirstWithHostNet | |
| - # Init containers are run only once in the lifetime of a pod, before | |
| - # it's started up for the first time. It has to exit successfully | |
| - # before the pod's main containers are allowed to start. | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # All addresses used to contact a node must be specified in the --addresses arg. | |
| - # | |
| - # In addition to the node certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=node -addresses=localhost,127.0.0.1,$(hostname -f),$(hostname -f|cut -f 1-2 -d '.'),cockroachdb-public,cockroachdb-public.$(hostname -f|cut -f 3- -d '.'),cockroachdb-public.$(hostname -f|cut -f 3-4 -d '.'),cockroachdb-public.$(hostname -f|cut -f 3 -d '.') -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: certs | |
| - mountPath: /cockroach-certs | |
| - # NOTE: If you are running clients that generate heavy load, you may find | |
| - # it useful to copy this anti-affinity policy into the client pods' | |
| - # configurations as well to avoid running them on the same machines as | |
| - # CockroachDB and interfering with each other's performance. | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - # NOTE: Always use the most recent version of CockroachDB for the best | |
| - # performance and reliability. | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - # TODO: Change these to appropriate values for the hardware that you're running. You can see | |
| - # the resources that can be allocated on each of your Kubernetes nodes by running: | |
| - # kubectl describe nodes | |
| - # Note that requests and limits should have identical values. | |
| - resources: | |
| - requests: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - limits: | |
| - cpu: "2" | |
| - memory: "8Gi" | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| -# We recommend that you do not configure a liveness probe on a production environment, as this can impact the availability of production databases. | |
| -# livenessProbe: | |
| -# httpGet: | |
| -# path: "/health" | |
| -# port: http | |
| -# scheme: HTTPS | |
| -# initialDelaySeconds: 30 | |
| -# periodSeconds: 5 | |
| - readinessProbe: | |
| - httpGet: | |
| - path: "/health?ready=1" | |
| - port: http | |
| - scheme: HTTPS | |
| - initialDelaySeconds: 10 | |
| - periodSeconds: 5 | |
| - failureThreshold: 2 | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - "exec /cockroach/cockroach start --logtostderr --certs-dir /cockroach/cockroach-certs --advertise-host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - emptyDir: {} | |
| - podManagementPolicy: Parallel | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - # TODO: This specifically asks for a storage class with the name "ssd". A | |
| - # storage class of this name doesn't exist by default. See our docs for | |
| - # more information on how to create an optimized storage class for use here: | |
| - # https://www.cockroachlabs.com/docs/stable/kubernetes-performance.html#disk-type | |
| - storageClassName: ssd | |
| - resources: | |
| - requests: | |
| - # TODO: This asks for a fairly large disk by default because on | |
| - # certain popular clouds there is a direct correlation between disk | |
| - # size and the IOPS provisioned to the disk. Change this as necessary | |
| - # to suit your needs, but be aware that smaller disks will typically | |
| - # mean worse performance. | |
| - storage: 1024Gi | |
| diff --git a/cloud/kubernetes/templates/v1.6/client-secure.yaml b/cloud/kubernetes/templates/v1.6/client-secure.yaml | |
| deleted file mode 100644 | |
| index 0df772c510..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.6/client-secure.yaml | |
| +++ /dev/null | |
| @@ -1,48 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Pod | |
| -metadata: | |
| - name: cockroachdb-client-secure | |
| - labels: | |
| - app: cockroachdb-client | |
| -spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cockroachdb-client | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - # Keep a pod open indefinitely so kubectl exec can be used to get a shell to it | |
| - # and run cockroach client commands, such as cockroach sql, cockroach node status, etc. | |
| - command: | |
| - - sleep | |
| - - "2147483648" # 2^31 | |
| - # This pod isn't doing anything important, so don't bother waiting to terminate it. | |
| - terminationGracePeriodSeconds: 0 | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/v1.6/cluster-init-secure.yaml b/cloud/kubernetes/templates/v1.6/cluster-init-secure.yaml | |
| deleted file mode 100644 | |
| index 830f2928a9..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.6/cluster-init-secure.yaml | |
| +++ /dev/null | |
| @@ -1,49 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init-secure | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--certs-dir=/cockroach-certs" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/v1.6/cluster-init.yaml b/cloud/kubernetes/templates/v1.6/cluster-init.yaml | |
| deleted file mode 100644 | |
| index 07a43d8456..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.6/cluster-init.yaml | |
| +++ /dev/null | |
| @@ -1,19 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--insecure" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| diff --git a/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset-secure.yaml | |
| deleted file mode 100644 | |
| index 178b870cb9..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,221 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # This is needed to make the peer-finder work properly and to help avoid | |
| - # edge cases where instance 0 comes up after losing its data and needs to | |
| - # decide whether it should create a new cluster or try to join an existing | |
| - # one. If it creates a new cluster when it should have joined an existing | |
| - # one, we'd end up with two separate clusters listening at the same service | |
| - # endpoint, which would be very bad. | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: apps/v1beta1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - # Init containers are run only once in the lifetime of a pod, before | |
| - # it's started up for the first time. It has to exit successfully | |
| - # before the pod's main containers are allowed to start. | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # All addresses used to contact a node must be specified in the --addresses arg. | |
| - # | |
| - # In addition to the node certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=node -addresses=localhost,127.0.0.1,$(hostname -f),$(hostname -f|cut -f 1-2 -d '.'),cockroachdb-public,cockroachdb-public.$(hostname -f|cut -f 3- -d '.'),cockroachdb-public.$(hostname -f|cut -f 3-4 -d '.'),cockroachdb-public.$(hostname -f|cut -f 3 -d '.') -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: certs | |
| - mountPath: /cockroach-certs | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - # Once 2.0 is out, we should be able to switch from --host to --advertise-host to make port-forwarding work to the main port. | |
| - - "exec /cockroach/cockroach start --logtostderr --certs-dir /cockroach/cockroach-certs --host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - emptyDir: {} | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset.yaml b/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset.yaml | |
| deleted file mode 100644 | |
| index 79379f4d37..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.6/cockroachdb-statefulset.yaml | |
| +++ /dev/null | |
| @@ -1,119 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # This is needed to make the peer-finder work properly and to help avoid | |
| - # edge cases where instance 0 comes up after losing its data and needs to | |
| - # decide whether it should create a new cluster or try to join an existing | |
| - # one. If it creates a new cluster when it should have joined an existing | |
| - # one, we'd end up with two separate clusters listening at the same service | |
| - # endpoint, which would be very bad. | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: apps/v1beta1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-insecure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - "exec /cockroach/cockroach start --logtostderr --insecure --advertise-host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/v1.7/client-secure.yaml b/cloud/kubernetes/templates/v1.7/client-secure.yaml | |
| deleted file mode 100644 | |
| index 0df772c510..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.7/client-secure.yaml | |
| +++ /dev/null | |
| @@ -1,48 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Pod | |
| -metadata: | |
| - name: cockroachdb-client-secure | |
| - labels: | |
| - app: cockroachdb-client | |
| -spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cockroachdb-client | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - # Keep a pod open indefinitely so kubectl exec can be used to get a shell to it | |
| - # and run cockroach client commands, such as cockroach sql, cockroach node status, etc. | |
| - command: | |
| - - sleep | |
| - - "2147483648" # 2^31 | |
| - # This pod isn't doing anything important, so don't bother waiting to terminate it. | |
| - terminationGracePeriodSeconds: 0 | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/v1.7/cluster-init-secure.yaml b/cloud/kubernetes/templates/v1.7/cluster-init-secure.yaml | |
| deleted file mode 100644 | |
| index 830f2928a9..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.7/cluster-init-secure.yaml | |
| +++ /dev/null | |
| @@ -1,49 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init-secure | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # In addition to the client certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=client -user=root -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - volumeMounts: | |
| - - name: client-certs | |
| - mountPath: /cockroach-certs | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--certs-dir=/cockroach-certs" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| - volumes: | |
| - - name: client-certs | |
| - emptyDir: {} | |
| diff --git a/cloud/kubernetes/templates/v1.7/cluster-init.yaml b/cloud/kubernetes/templates/v1.7/cluster-init.yaml | |
| deleted file mode 100644 | |
| index 07a43d8456..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.7/cluster-init.yaml | |
| +++ /dev/null | |
| @@ -1,19 +0,0 @@ | |
| -apiVersion: batch/v1 | |
| -kind: Job | |
| -metadata: | |
| - name: cluster-init | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - template: | |
| - spec: | |
| - containers: | |
| - - name: cluster-init | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/cockroach/cockroach" | |
| - - "init" | |
| - - "--insecure" | |
| - - "--host=cockroachdb-0.cockroachdb" | |
| - restartPolicy: OnFailure | |
| diff --git a/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset-secure.yaml | |
| deleted file mode 100644 | |
| index 46c278ae95..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset-secure.yaml | |
| +++ /dev/null | |
| @@ -1,233 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: ServiceAccount | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: Role | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - "" | |
| - resources: | |
| - - secrets | |
| - verbs: | |
| - - create | |
| - - get | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: ClusterRole | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -rules: | |
| -- apiGroups: | |
| - - certificates.k8s.io | |
| - resources: | |
| - - certificatesigningrequests | |
| - verbs: | |
| - - create | |
| - - get | |
| - - watch | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: RoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: Role | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: rbac.authorization.k8s.io/v1beta1 | |
| -kind: ClusterRoleBinding | |
| -metadata: | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| -roleRef: | |
| - apiGroup: rbac.authorization.k8s.io | |
| - kind: ClusterRole | |
| - name: cockroachdb | |
| -subjects: | |
| -- kind: ServiceAccount | |
| - name: cockroachdb | |
| - namespace: default | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # This is needed to make the peer-finder work properly and to help avoid | |
| - # edge cases where instance 0 comes up after losing its data and needs to | |
| - # decide whether it should create a new cluster or try to join an existing | |
| - # one. If it creates a new cluster when it should have joined an existing | |
| - # one, we'd end up with two separate clusters listening at the same service | |
| - # endpoint, which would be very bad. | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1beta1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - serviceAccountName: cockroachdb | |
| - # Init containers are run only once in the lifetime of a pod, before | |
| - # it's started up for the first time. It has to exit successfully | |
| - # before the pod's main containers are allowed to start. | |
| - initContainers: | |
| - # The init-certs container sends a certificate signing request to the | |
| - # kubernetes cluster. | |
| - # You can see pending requests using: kubectl get csr | |
| - # CSRs can be approved using: kubectl certificate approve <csr name> | |
| - # | |
| - # All addresses used to contact a node must be specified in the --addresses arg. | |
| - # | |
| - # In addition to the node certificate and key, the init-certs entrypoint will symlink | |
| - # the cluster CA to the certs directory. | |
| - - name: init-certs | |
| - image: cockroachdb/cockroach-k8s-request-cert:0.4 | |
| - imagePullPolicy: IfNotPresent | |
| - command: | |
| - - "/bin/ash" | |
| - - "-ecx" | |
| - - "/request-cert -namespace=${POD_NAMESPACE} -certs-dir=/cockroach-certs -type=node -addresses=localhost,127.0.0.1,$(hostname -f),$(hostname -f|cut -f 1-2 -d '.'),cockroachdb-public,cockroachdb-public.$(hostname -f|cut -f 3- -d '.'),cockroachdb-public.$(hostname -f|cut -f 3-4 -d '.'),cockroachdb-public.$(hostname -f|cut -f 3 -d '.') -symlink-ca-from=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt" | |
| - env: | |
| - - name: POD_NAMESPACE | |
| - valueFrom: | |
| - fieldRef: | |
| - fieldPath: metadata.namespace | |
| - volumeMounts: | |
| - - name: certs | |
| - mountPath: /cockroach-certs | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - - name: certs | |
| - mountPath: /cockroach/cockroach-certs | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-secure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - # Once 2.0 is out, we should be able to switch from --host to --advertise-host to make port-forwarding work to the main port. | |
| - - "exec /cockroach/cockroach start --logtostderr --certs-dir /cockroach/cockroach-certs --host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - - name: certs | |
| - emptyDir: {} | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset.yaml b/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset.yaml | |
| deleted file mode 100644 | |
| index 6cdb744279..0000000000 | |
| --- a/cloud/kubernetes/templates/v1.7/cockroachdb-statefulset.yaml | |
| +++ /dev/null | |
| @@ -1,131 +0,0 @@ | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service is meant to be used by clients of the database. It exposes a ClusterIP that will | |
| - # automatically load balance connections to the different database pods. | |
| - name: cockroachdb-public | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - ports: | |
| - # The main port, served by gRPC, serves Postgres-flavor SQL, internode | |
| - # traffic and the cli. | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - # The secondary port serves the UI as well as health and debug endpoints. | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: v1 | |
| -kind: Service | |
| -metadata: | |
| - # This service only exists to create DNS entries for each pod in the stateful | |
| - # set such that they can resolve each other's IP addresses. It does not | |
| - # create a load-balanced ClusterIP and should not be used directly by clients | |
| - # in most circumstances. | |
| - name: cockroachdb | |
| - labels: | |
| - app: cockroachdb | |
| - annotations: | |
| - # This is needed to make the peer-finder work properly and to help avoid | |
| - # edge cases where instance 0 comes up after losing its data and needs to | |
| - # decide whether it should create a new cluster or try to join an existing | |
| - # one. If it creates a new cluster when it should have joined an existing | |
| - # one, we'd end up with two separate clusters listening at the same service | |
| - # endpoint, which would be very bad. | |
| - service.alpha.kubernetes.io/tolerate-unready-endpoints: "true" | |
| - # Enable automatic monitoring of all instances when Prometheus is running in the cluster. | |
| - prometheus.io/scrape: "true" | |
| - prometheus.io/path: "_status/vars" | |
| - prometheus.io/port: "8080" | |
| -spec: | |
| - ports: | |
| - - port: 26257 | |
| - targetPort: 26257 | |
| - name: grpc | |
| - - port: 8080 | |
| - targetPort: 8080 | |
| - name: http | |
| - clusterIP: None | |
| - selector: | |
| - app: cockroachdb | |
| ---- | |
| -apiVersion: policy/v1beta1 | |
| -kind: PodDisruptionBudget | |
| -metadata: | |
| - name: cockroachdb-budget | |
| - labels: | |
| - app: cockroachdb | |
| -spec: | |
| - selector: | |
| - matchLabels: | |
| - app: cockroachdb | |
| - maxUnavailable: 1 | |
| ---- | |
| -apiVersion: apps/v1beta1 | |
| -kind: StatefulSet | |
| -metadata: | |
| - name: cockroachdb | |
| -spec: | |
| - serviceName: "cockroachdb" | |
| - replicas: 3 | |
| - template: | |
| - metadata: | |
| - labels: | |
| - app: cockroachdb | |
| - spec: | |
| - affinity: | |
| - podAntiAffinity: | |
| - preferredDuringSchedulingIgnoredDuringExecution: | |
| - - weight: 100 | |
| - podAffinityTerm: | |
| - labelSelector: | |
| - matchExpressions: | |
| - - key: app | |
| - operator: In | |
| - values: | |
| - - cockroachdb | |
| - topologyKey: kubernetes.io/hostname | |
| - containers: | |
| - - name: cockroachdb | |
| - image: cockroachdb/cockroach:@VERSION@ | |
| - imagePullPolicy: IfNotPresent | |
| - ports: | |
| - - containerPort: 26257 | |
| - name: grpc | |
| - - containerPort: 8080 | |
| - name: http | |
| - volumeMounts: | |
| - - name: datadir | |
| - mountPath: /cockroach/cockroach-data | |
| - env: | |
| - - name: COCKROACH_CHANNEL | |
| - value: kubernetes-insecure | |
| - command: | |
| - - "/bin/bash" | |
| - - "-ecx" | |
| - # The use of qualified `hostname -f` is crucial: | |
| - # Other nodes aren't able to look up the unqualified hostname. | |
| - - "exec /cockroach/cockroach start --logtostderr --insecure --advertise-host $(hostname -f) --http-addr 0.0.0.0 --join cockroachdb-0.cockroachdb,cockroachdb-1.cockroachdb,cockroachdb-2.cockroachdb --cache 25% --max-sql-memory 25%" | |
| - # No pre-stop hook is required, a SIGTERM plus some time is all that's | |
| - # needed for graceful shutdown of a node. | |
| - terminationGracePeriodSeconds: 60 | |
| - volumes: | |
| - - name: datadir | |
| - persistentVolumeClaim: | |
| - claimName: datadir | |
| - updateStrategy: | |
| - type: RollingUpdate | |
| - volumeClaimTemplates: | |
| - - metadata: | |
| - name: datadir | |
| - spec: | |
| - accessModes: | |
| - - "ReadWriteOnce" | |
| - resources: | |
| - requests: | |
| - storage: 100Gi | |
| diff --git a/cloud/kubernetes/v1.6/client-secure.yaml b/cloud/kubernetes/v1.6/client-secure.yaml | |
| index 58c9105972..d0c7967dbd 100644 | |
| --- a/cloud/kubernetes/v1.6/client-secure.yaml | |
| +++ b/cloud/kubernetes/v1.6/client-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.6/client-secure.yaml | |
| apiVersion: v1 | |
| kind: Pod | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.6/cluster-init-secure.yaml b/cloud/kubernetes/v1.6/cluster-init-secure.yaml | |
| index b17c6a2bd6..0154e8680e 100644 | |
| --- a/cloud/kubernetes/v1.6/cluster-init-secure.yaml | |
| +++ b/cloud/kubernetes/v1.6/cluster-init-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.6/cluster-init-secure.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.6/cluster-init.yaml b/cloud/kubernetes/v1.6/cluster-init.yaml | |
| index d323511fc9..d33a6fb7d0 100644 | |
| --- a/cloud/kubernetes/v1.6/cluster-init.yaml | |
| +++ b/cloud/kubernetes/v1.6/cluster-init.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.6/cluster-init.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.6/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/v1.6/cockroachdb-statefulset-secure.yaml | |
| index a9110e5235..b269400429 100644 | |
| --- a/cloud/kubernetes/v1.6/cockroachdb-statefulset-secure.yaml | |
| +++ b/cloud/kubernetes/v1.6/cockroachdb-statefulset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.6/cockroachdb-statefulset-secure.yaml | |
| apiVersion: v1 | |
| kind: ServiceAccount | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.6/cockroachdb-statefulset.yaml b/cloud/kubernetes/v1.6/cockroachdb-statefulset.yaml | |
| index 42712a7bca..2de7761fef 100644 | |
| --- a/cloud/kubernetes/v1.6/cockroachdb-statefulset.yaml | |
| +++ b/cloud/kubernetes/v1.6/cockroachdb-statefulset.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.6/cockroachdb-statefulset.yaml | |
| apiVersion: v1 | |
| kind: Service | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.7/client-secure.yaml b/cloud/kubernetes/v1.7/client-secure.yaml | |
| index 9b94872ed2..d0c7967dbd 100644 | |
| --- a/cloud/kubernetes/v1.7/client-secure.yaml | |
| +++ b/cloud/kubernetes/v1.7/client-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.7/client-secure.yaml | |
| apiVersion: v1 | |
| kind: Pod | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.7/cluster-init-secure.yaml b/cloud/kubernetes/v1.7/cluster-init-secure.yaml | |
| index ca43db7558..0154e8680e 100644 | |
| --- a/cloud/kubernetes/v1.7/cluster-init-secure.yaml | |
| +++ b/cloud/kubernetes/v1.7/cluster-init-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.7/cluster-init-secure.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.7/cluster-init.yaml b/cloud/kubernetes/v1.7/cluster-init.yaml | |
| index 5fde6df11d..d33a6fb7d0 100644 | |
| --- a/cloud/kubernetes/v1.7/cluster-init.yaml | |
| +++ b/cloud/kubernetes/v1.7/cluster-init.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.7/cluster-init.yaml | |
| apiVersion: batch/v1 | |
| kind: Job | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.7/cockroachdb-statefulset-secure.yaml b/cloud/kubernetes/v1.7/cockroachdb-statefulset-secure.yaml | |
| index f10d589062..1b682f8c9e 100644 | |
| --- a/cloud/kubernetes/v1.7/cockroachdb-statefulset-secure.yaml | |
| +++ b/cloud/kubernetes/v1.7/cockroachdb-statefulset-secure.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.7/cockroachdb-statefulset-secure.yaml | |
| apiVersion: v1 | |
| kind: ServiceAccount | |
| metadata: | |
| diff --git a/cloud/kubernetes/v1.7/cockroachdb-statefulset.yaml b/cloud/kubernetes/v1.7/cockroachdb-statefulset.yaml | |
| index 8be582972a..26937db23c 100644 | |
| --- a/cloud/kubernetes/v1.7/cockroachdb-statefulset.yaml | |
| +++ b/cloud/kubernetes/v1.7/cockroachdb-statefulset.yaml | |
| @@ -1,4 +1,3 @@ | |
| -# Generated file, DO NOT EDIT. Source: cloud/kubernetes/templates/v1.7/cockroachdb-statefulset.yaml | |
| apiVersion: v1 | |
| kind: Service | |
| metadata: | |
| diff --git a/dev b/dev | |
| index 8b2d7d91b1..410066c5ad 100755 | |
| --- a/dev | |
| +++ b/dev | |
| @@ -3,7 +3,7 @@ | |
| set -euo pipefail | |
| # Bump this counter to force rebuilding `dev` on all machines. | |
| -DEV_VERSION=20 | |
| +DEV_VERSION=19 | |
| THIS_DIR=$(cd "$(dirname "$0")" && pwd) | |
| BINARY_DIR=$THIS_DIR/bin/dev-versions | |
| diff --git a/docs/generated/eventlog.md b/docs/generated/eventlog.md | |
| index cc042a51e9..15969ca0ff 100644 | |
| --- a/docs/generated/eventlog.md | |
| +++ b/docs/generated/eventlog.md | |
| @@ -2305,32 +2305,6 @@ are automatically converted server-side. | |
| Events in this category are logged to the `TELEMETRY` channel. | |
| -### `captured_index_usage_stats` | |
| - | |
| -An event of type `captured_index_usage_stats` | |
| - | |
| - | |
| -| Field | Description | Sensitive | | |
| -|--|--|--| | |
| -| `TotalReadCount` | TotalReadCount is the number of times this index has been read from. | no | | |
| -| `LastRead` | LastRead is the timestamp that this index was last being read from. | yes | | |
| -| `TableID` | TableID is the ID of the table this index is created on. This is same as descpb.TableID and is unique within the cluster. | no | | |
| -| `IndexID` | IndexID is the ID of the index within the scope of the given table. | no | | |
| -| `DatabaseName` | | yes | | |
| -| `TableName` | | yes | | |
| -| `IndexName` | | yes | | |
| -| `IndexType` | | yes | | |
| -| `IsUnique` | | no | | |
| -| `IsInverted` | | no | | |
| - | |
| - | |
| -#### Common fields | |
| - | |
| -| Field | Description | Sensitive | | |
| -|--|--|--| | |
| -| `Timestamp` | The timestamp of the event. Expressed as nanoseconds since the Unix epoch. | no | | |
| -| `EventType` | The type of the event. | no | | |
| - | |
| ### `sampled_query` | |
| An event of type `sampled_query` is the SQL query event logged to the telemetry channel. It | |
| diff --git a/docs/generated/http/BUILD.bazel b/docs/generated/http/BUILD.bazel | |
| index 5e5ac861e7..f171a56bfd 100644 | |
| --- a/docs/generated/http/BUILD.bazel | |
| +++ b/docs/generated/http/BUILD.bazel | |
| @@ -1,6 +1,19 @@ | |
| genrule( | |
| name = "http", | |
| srcs = [ | |
| + "//pkg/ts/tspb:tspb_proto", | |
| + "//pkg/util/metric:metric_proto", | |
| + "//pkg/ts/catalog:catalog_proto", | |
| + "//pkg/util/duration:duration_proto", | |
| + "//pkg/util/timeutil/pgdate:pgdate_proto", | |
| + "//pkg/sql/sessiondatapb:sessiondatapb_proto", | |
| + "//pkg/sql/inverted:inverted_proto", | |
| + "//pkg/sql/pgwire/pgerror:pgerror_proto", | |
| + "//pkg/sql/contentionpb:contentionpb_proto", | |
| + "//pkg/sql/execinfrapb:execinfrapb_proto", | |
| + "//pkg/kv/kvserver/kvserverpb:kvserverpb_proto", | |
| + "//pkg/kv/kvserver/liveness/livenesspb:livenesspb_proto", | |
| + "//pkg/util/log/logpb:logpb_proto", | |
| "//pkg/build:build_proto", | |
| "//pkg/clusterversion:clusterversion_proto", | |
| "//pkg/config/zonepb:zonepb_proto", | |
| @@ -9,8 +22,6 @@ genrule( | |
| "//pkg/gossip:gossip_proto", | |
| "//pkg/jobs/jobspb:jobspb_proto", | |
| "//pkg/kv/kvserver/concurrency/lock:lock_proto", | |
| - "//pkg/kv/kvserver/kvserverpb:kvserverpb_proto", | |
| - "//pkg/kv/kvserver/liveness/livenesspb:livenesspb_proto", | |
| "//pkg/kv/kvserver/readsummary/rspb:rspb_proto", | |
| "//pkg/roachpb:roachpb_proto", | |
| "//pkg/server/diagnostics/diagnosticspb:diagnosticspb_proto", | |
| @@ -19,24 +30,12 @@ genrule( | |
| "//pkg/settings:settings_proto", | |
| "//pkg/sql/catalog/catpb:catpb_proto", | |
| "//pkg/sql/catalog/descpb:descpb_proto", | |
| - "//pkg/sql/contentionpb:contentionpb_proto", | |
| - "//pkg/sql/execinfrapb:execinfrapb_proto", | |
| - "//pkg/sql/inverted:inverted_proto", | |
| - "//pkg/sql/lex:lex_proto", | |
| - "//pkg/sql/pgwire/pgerror:pgerror_proto", | |
| "//pkg/sql/schemachanger/scpb:scpb_proto", | |
| - "//pkg/sql/sessiondatapb:sessiondatapb_proto", | |
| "//pkg/sql/types:types_proto", | |
| "//pkg/storage/enginepb:enginepb_proto", | |
| - "//pkg/ts/catalog:catalog_proto", | |
| - "//pkg/ts/tspb:tspb_proto", | |
| - "//pkg/util/duration:duration_proto", | |
| + "//pkg/util:util_proto", | |
| "//pkg/util/hlc:hlc_proto", | |
| - "//pkg/util/log/logpb:logpb_proto", | |
| - "//pkg/util/metric:metric_proto", | |
| - "//pkg/util/timeutil/pgdate:pgdate_proto", | |
| "//pkg/util/tracing/tracingpb:tracingpb_proto", | |
| - "//pkg/util:util_proto", | |
| "@com_github_prometheus_client_model//io/prometheus/client:client_proto", | |
| "@com_github_cockroachdb_errors//errorspb:errorspb_proto", | |
| "@com_github_gogo_protobuf//gogoproto:gogo_proto", | |
| diff --git a/docs/generated/http/full.md b/docs/generated/http/full.md | |
| index 5597f184fd..3595e3437d 100644 | |
| --- a/docs/generated/http/full.md | |
| +++ b/docs/generated/http/full.md | |
| @@ -1229,9 +1229,6 @@ Support status: [reserved](#support-status) | |
| | locks_with_wait_queues | [int64](#cockroach.server.serverpb.RaftDebugResponse-int64) | | | [reserved](#support-status) | | |
| | lock_wait_queue_waiters | [int64](#cockroach.server.serverpb.RaftDebugResponse-int64) | | | [reserved](#support-status) | | |
| | top_k_locks_by_wait_queue_waiters | [RangeInfo.LockInfo](#cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.RangeInfo.LockInfo) | repeated | | [reserved](#support-status) | | |
| -| locality | [Locality](#cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.Locality) | | | [reserved](#support-status) | | |
| -| is_leaseholder | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) | | | [reserved](#support-status) | | |
| -| lease_valid | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) | | | [reserved](#support-status) | | |
| @@ -1356,34 +1353,6 @@ only. | |
| -<a name="cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.Locality"></a> | |
| -#### Locality | |
| - | |
| -Locality is an ordered set of key value Tiers that describe a node's | |
| -location. The tier keys should be the same across all nodes. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| tiers | [Tier](#cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.Tier) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.Tier"></a> | |
| -#### Tier | |
| - | |
| -Tier represents one level of the locality hierarchy. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [string](#cockroach.server.serverpb.RaftDebugResponse-string) | | Key is the name of tier and should match all other nodes. | [reserved](#support-status) | | |
| -| value | [string](#cockroach.server.serverpb.RaftDebugResponse-string) | | Value is node specific value corresponding to the key. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| <a name="cockroach.server.serverpb.RaftDebugResponse-cockroach.server.serverpb.RaftRangeError"></a> | |
| #### RaftRangeError | |
| @@ -1467,9 +1436,6 @@ Support status: [reserved](#support-status) | |
| | locks_with_wait_queues | [int64](#cockroach.server.serverpb.RangesResponse-int64) | | | [reserved](#support-status) | | |
| | lock_wait_queue_waiters | [int64](#cockroach.server.serverpb.RangesResponse-int64) | | | [reserved](#support-status) | | |
| | top_k_locks_by_wait_queue_waiters | [RangeInfo.LockInfo](#cockroach.server.serverpb.RangesResponse-cockroach.server.serverpb.RangeInfo.LockInfo) | repeated | | [reserved](#support-status) | | |
| -| locality | [Locality](#cockroach.server.serverpb.RangesResponse-cockroach.server.serverpb.Locality) | | | [reserved](#support-status) | | |
| -| is_leaseholder | [bool](#cockroach.server.serverpb.RangesResponse-bool) | | | [reserved](#support-status) | | |
| -| lease_valid | [bool](#cockroach.server.serverpb.RangesResponse-bool) | | | [reserved](#support-status) | | |
| @@ -1594,206 +1560,6 @@ only. | |
| -<a name="cockroach.server.serverpb.RangesResponse-cockroach.server.serverpb.Locality"></a> | |
| -#### Locality | |
| - | |
| -Locality is an ordered set of key value Tiers that describe a node's | |
| -location. The tier keys should be the same across all nodes. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| tiers | [Tier](#cockroach.server.serverpb.RangesResponse-cockroach.server.serverpb.Tier) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.RangesResponse-cockroach.server.serverpb.Tier"></a> | |
| -#### Tier | |
| - | |
| -Tier represents one level of the locality hierarchy. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [string](#cockroach.server.serverpb.RangesResponse-string) | | Key is the name of tier and should match all other nodes. | [reserved](#support-status) | | |
| -| value | [string](#cockroach.server.serverpb.RangesResponse-string) | | Value is node specific value corresponding to the key. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -## TenantRanges | |
| - | |
| -`GET /_status/tenant_ranges` | |
| - | |
| -TenantRanges requests internal details about all range replicas within | |
| -the tenant's keyspace. | |
| - | |
| -Support status: [reserved](#support-status) | |
| - | |
| -#### Request Parameters | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -#### Response Parameters | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| ranges_by_locality | [TenantRangesResponse.RangesByLocalityEntry](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangesResponse.RangesByLocalityEntry) | repeated | ranges_by_locality maps each range replica to its specified availability zone, as defined within the replica's locality metadata (default key `az`). Replicas without the default available zone key set will fall under the `locality-unset` key. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangesResponse.RangesByLocalityEntry"></a> | |
| -#### TenantRangesResponse.RangesByLocalityEntry | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | | | | |
| -| value | [TenantRangesResponse.TenantRangeList](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangesResponse.TenantRangeList) | | | | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangesResponse.TenantRangeList"></a> | |
| -#### TenantRangesResponse.TenantRangeList | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| ranges | [TenantRangeInfo](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangeInfo) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangeInfo"></a> | |
| -#### TenantRangeInfo | |
| - | |
| -TenantRangeInfo provides metadata about a specific range replica, | |
| -where concepts not considered to be relevant within the tenant | |
| -abstraction (e.g. NodeIDs) are omitted. Instead, Locality information | |
| -is used to distinguish replicas. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| range_id | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The ID of the Range. | [reserved](#support-status) | | |
| -| span | [PrettySpan](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.PrettySpan) | | The pretty-printed key span of the range. | [reserved](#support-status) | | |
| -| locality | [Locality](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.Locality) | | Any locality information associated with this specific replica. | [reserved](#support-status) | | |
| -| is_leaseholder | [bool](#cockroach.server.serverpb.TenantRangesResponse-bool) | | Whether the range's specific replica is a leaseholder. | [reserved](#support-status) | | |
| -| lease_valid | [bool](#cockroach.server.serverpb.TenantRangesResponse-bool) | | Whether the range's specific replica holds a valid lease. | [reserved](#support-status) | | |
| -| range_stats | [RangeStatistics](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.RangeStatistics) | | Statistics about the range replica, e.g. QPS, WPS. | [reserved](#support-status) | | |
| -| mvcc_stats | [cockroach.storage.enginepb.MVCCStats](#cockroach.server.serverpb.TenantRangesResponse-cockroach.storage.enginepb.MVCCStats) | | MVCC stats about the range replica, e.g. live_bytes. | [reserved](#support-status) | | |
| -| read_latches | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | Read count as reported by the range replica's spanlatch.Manager. | [reserved](#support-status) | | |
| -| write_latches | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | Write count as reported by the range replica's spanlatch.Manager. | [reserved](#support-status) | | |
| -| locks | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The number of locks as reported by the range replica's lockTable. | [reserved](#support-status) | | |
| -| locks_with_wait_queues | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The number of locks with non-empty wait-queues as reported by the range replica's lockTable | [reserved](#support-status) | | |
| -| lock_wait_queue_waiters | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The aggregate number of waiters in wait-queues across all locks as reported by the range replica's lockTable | [reserved](#support-status) | | |
| -| top_k_locks_by_wait_queue_waiters | [TenantRangeInfo.LockInfo](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangeInfo.LockInfo) | repeated | The top-k locks with the most waiters (readers + writers) in their wait-queue, ordered in descending order. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.PrettySpan"></a> | |
| -#### PrettySpan | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| start_key | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | | [reserved](#support-status) | | |
| -| end_key | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.Locality"></a> | |
| -#### Locality | |
| - | |
| -Locality is an ordered set of key value Tiers that describe a node's | |
| -location. The tier keys should be the same across all nodes. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| tiers | [Tier](#cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.Tier) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.Tier"></a> | |
| -#### Tier | |
| - | |
| -Tier represents one level of the locality hierarchy. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | Key is the name of tier and should match all other nodes. | [reserved](#support-status) | | |
| -| value | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | Value is node specific value corresponding to the key. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.RangeStatistics"></a> | |
| -#### RangeStatistics | |
| - | |
| -RangeStatistics describes statistics reported by a range. For internal use | |
| -only. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| queries_per_second | [double](#cockroach.server.serverpb.TenantRangesResponse-double) | | Queries per second served by this range.<br><br>Note that queries per second will only be known by the leaseholder. All other replicas will report it as 0. | [reserved](#support-status) | | |
| -| writes_per_second | [double](#cockroach.server.serverpb.TenantRangesResponse-double) | | Writes per second served by this range. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.TenantRangesResponse-cockroach.server.serverpb.TenantRangeInfo.LockInfo"></a> | |
| -#### TenantRangeInfo.LockInfo | |
| - | |
| -LockInfo provides metadata about the state of a single lock | |
| -in the range replica's lockTable. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| pretty_key | [string](#cockroach.server.serverpb.TenantRangesResponse-string) | | The lock's key in pretty format. | [reserved](#support-status) | | |
| -| key | [bytes](#cockroach.server.serverpb.TenantRangesResponse-bytes) | | The lock's key. | [reserved](#support-status) | | |
| -| held | [bool](#cockroach.server.serverpb.TenantRangesResponse-bool) | | Is the lock actively held by a transaction, or just a reservation? | [reserved](#support-status) | | |
| -| waiters | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The number of waiters in the lock's wait queue. | [reserved](#support-status) | | |
| -| waiting_readers | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The number of waiting readers in the lock's wait queue. | [reserved](#support-status) | | |
| -| waiting_writers | [int64](#cockroach.server.serverpb.TenantRangesResponse-int64) | | The number of waiting writers in the lock's wait queue. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| ## Gossip | |
| @@ -3364,8 +3130,6 @@ of ranges currently considered “hot” by the node(s). | |
| | Field | Type | Label | Description | Support status | | |
| | ----- | ---- | ----- | ----------- | -------------- | | |
| | node_id | [string](#cockroach.server.serverpb.HotRangesRequest-string) | | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) | | |
| -| page_size | [int32](#cockroach.server.serverpb.HotRangesRequest-int32) | | | [reserved](#support-status) | | |
| -| page_token | [string](#cockroach.server.serverpb.HotRangesRequest-string) | | | [reserved](#support-status) | | |
| @@ -3445,92 +3209,6 @@ target node(s) selected in a HotRangesRequest. | |
| | ----- | ---- | ----- | ----------- | -------------- | | |
| | desc | [cockroach.roachpb.RangeDescriptor](#cockroach.server.serverpb.HotRangesResponse-cockroach.roachpb.RangeDescriptor) | | Desc is the descriptor of the range for which the report was produced.<br><br>TODO(knz): This field should be removed. See: https://github.com/cockroachdb/cockroach/issues/53212 | [reserved](#support-status) | | |
| | queries_per_second | [double](#cockroach.server.serverpb.HotRangesResponse-double) | | QueriesPerSecond is the recent number of queries per second on this range. | [alpha](#support-status) | | |
| -| leaseholder_node_id | [int32](#cockroach.server.serverpb.HotRangesResponse-int32) | | LeaseholderNodeID indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -## HotRangesV2 | |
| - | |
| -`POST /_status/v2/hotranges` | |
| - | |
| - | |
| - | |
| -Support status: [reserved](#support-status) | |
| - | |
| -#### Request Parameters | |
| - | |
| - | |
| - | |
| - | |
| -HotRangesRequest queries one or more cluster nodes for a list | |
| -of ranges currently considered “hot” by the node(s). | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| node_id | [string](#cockroach.server.serverpb.HotRangesRequest-string) | | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) | | |
| -| page_size | [int32](#cockroach.server.serverpb.HotRangesRequest-int32) | | | [reserved](#support-status) | | |
| -| page_token | [string](#cockroach.server.serverpb.HotRangesRequest-string) | | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -#### Response Parameters | |
| - | |
| - | |
| - | |
| - | |
| -HotRangesResponseV2 is a response payload returned by `HotRangesV2` service. | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| ranges | [HotRangesResponseV2.HotRange](#cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.HotRange) | repeated | Ranges contain list of hot ranges info that has highest number of QPS. | [reserved](#support-status) | | |
| -| errors_by_node_id | [HotRangesResponseV2.ErrorsByNodeIdEntry](#cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.ErrorsByNodeIdEntry) | repeated | errors contains any errors that occurred during fan-out calls to other nodes. | [reserved](#support-status) | | |
| -| next_page_token | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | NextPageToken represents next pagination token to request next slice of data. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.HotRange"></a> | |
| -#### HotRangesResponseV2.HotRange | |
| - | |
| -HotRange message describes a single hot range, ie its QPS, node ID it belongs to, etc. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| range_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | | range_id indicates Range ID that's identified as hot range. | [reserved](#support-status) | | |
| -| node_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | | node_id indicates the node that contains the current hot range. | [reserved](#support-status) | | |
| -| qps | [double](#cockroach.server.serverpb.HotRangesResponseV2-double) | | qps (queries per second) shows the amount of queries that interact with current range. | [reserved](#support-status) | | |
| -| table_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | table_name indicates the SQL table that the range belongs to. | [reserved](#support-status) | | |
| -| database_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | database_name indicates on database that has current hot range. | [reserved](#support-status) | | |
| -| index_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | index_name indicates the index name for current range. | [reserved](#support-status) | | |
| -| replica_node_ids | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | repeated | replica_node_ids specifies the list of node ids that contain replicas with current hot range. | [reserved](#support-status) | | |
| -| leaseholder_node_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | | leaseholder_node_id indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) | | |
| -| schema_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | schema_name provides the name of schema (if exists) for table in current range. | [reserved](#support-status) | | |
| -| store_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | | store_id indicates the Store ID where range is stored. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.ErrorsByNodeIdEntry"></a> | |
| -#### HotRangesResponseV2.ErrorsByNodeIdEntry | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | | | | | |
| -| value | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) | | | | | |
| @@ -3636,9 +3314,6 @@ Support status: [reserved](#support-status) | |
| | locks_with_wait_queues | [int64](#cockroach.server.serverpb.RangeResponse-int64) | | | [reserved](#support-status) | | |
| | lock_wait_queue_waiters | [int64](#cockroach.server.serverpb.RangeResponse-int64) | | | [reserved](#support-status) | | |
| | top_k_locks_by_wait_queue_waiters | [RangeInfo.LockInfo](#cockroach.server.serverpb.RangeResponse-cockroach.server.serverpb.RangeInfo.LockInfo) | repeated | | [reserved](#support-status) | | |
| -| locality | [Locality](#cockroach.server.serverpb.RangeResponse-cockroach.server.serverpb.Locality) | | | [reserved](#support-status) | | |
| -| is_leaseholder | [bool](#cockroach.server.serverpb.RangeResponse-bool) | | | [reserved](#support-status) | | |
| -| lease_valid | [bool](#cockroach.server.serverpb.RangeResponse-bool) | | | [reserved](#support-status) | | |
| @@ -3763,34 +3438,6 @@ only. | |
| -<a name="cockroach.server.serverpb.RangeResponse-cockroach.server.serverpb.Locality"></a> | |
| -#### Locality | |
| - | |
| -Locality is an ordered set of key value Tiers that describe a node's | |
| -location. The tier keys should be the same across all nodes. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| tiers | [Tier](#cockroach.server.serverpb.RangeResponse-cockroach.server.serverpb.Tier) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -<a name="cockroach.server.serverpb.RangeResponse-cockroach.server.serverpb.Tier"></a> | |
| -#### Tier | |
| - | |
| -Tier represents one level of the locality hierarchy. | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| key | [string](#cockroach.server.serverpb.RangeResponse-string) | | Key is the name of tier and should match all other nodes. | [reserved](#support-status) | | |
| -| value | [string](#cockroach.server.serverpb.RangeResponse-string) | | Value is node specific value corresponding to the key. | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| ## Diagnostics | |
| @@ -4793,51 +4440,6 @@ Response object for issuing Transaction ID Resolution. | |
| -## TransactionContentionEvents | |
| - | |
| -`GET /_status/transactioncontentionevents` | |
| - | |
| -TransactionContentionEvents returns a list of un-aggregated contention | |
| -events sorted by the collection timestamp. | |
| - | |
| -Support status: [reserved](#support-status) | |
| - | |
| -#### Request Parameters | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| node_id | [string](#cockroach.server.serverpb.TransactionContentionEventsRequest-string) | | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -#### Response Parameters | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| -| Field | Type | Label | Description | Support status | | |
| -| ----- | ---- | ----- | ----------- | -------------- | | |
| -| events | [cockroach.sql.contentionpb.ExtendedContentionEvent](#cockroach.server.serverpb.TransactionContentionEventsResponse-cockroach.sql.contentionpb.ExtendedContentionEvent) | repeated | | [reserved](#support-status) | | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - | |
| ## RequestCA | |
| `GET /_join/v1/ca` | |
| diff --git a/docs/generated/http/hotranges-other.md b/docs/generated/http/hotranges-other.md | |
| index fa43e6f7b9..55397521a6 100644 | |
| --- a/docs/generated/http/hotranges-other.md | |
| +++ b/docs/generated/http/hotranges-other.md | |
| @@ -62,6 +62,5 @@ Support status: [alpha](#support-status) | |
| | ----- | ---- | ----- | ----------- | -------------- | | |
| | desc | [cockroach.roachpb.RangeDescriptor](#cockroach.roachpb.RangeDescriptor) | | Desc is the descriptor of the range for which the report was produced.<br><br>TODO(knz): This field should be removed. See: https://github.com/cockroachdb/cockroach/issues/53212 | [reserved](#support-status) | | |
| | queries_per_second | [double](#double) | | QueriesPerSecond is the recent number of queries per second on this range. | [alpha](#support-status) | | |
| -| leaseholder_node_id | [int32](#int32) | | LeaseholderNodeID indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) | | |
| diff --git a/docs/generated/http/hotranges-request.md b/docs/generated/http/hotranges-request.md | |
| index c7ccfd28d4..6c542161d1 100644 | |
| --- a/docs/generated/http/hotranges-request.md | |
| +++ b/docs/generated/http/hotranges-request.md | |
| @@ -12,7 +12,5 @@ Support status: [alpha](#support-status) | |
| | Field | Type | Label | Description | Support status | | |
| | ----- | ---- | ----- | ----------- | -------------- | | |
| | node_id | [string](#string) | | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) | | |
| -| page_size | [int32](#int32) | | | [reserved](#support-status) | | |
| -| page_token | [string](#string) | | | [reserved](#support-status) | | |
| diff --git a/docs/generated/redact_safe.md b/docs/generated/redact_safe.md | |
| index 61ebfa5f1f..d91bd73b2b 100644 | |
| --- a/docs/generated/redact_safe.md | |
| +++ b/docs/generated/redact_safe.md | |
| @@ -7,13 +7,8 @@ pkg/base/node_id.go | `*SQLIDContainer` | |
| pkg/base/node_id.go | `*StoreIDContainer` | |
| pkg/cli/exit/exit.go | `Code` | |
| pkg/jobs/jobspb/wrap.go | `Type` | |
| -pkg/kv/bulk/sst_batcher.go | `sorted` | |
| -pkg/kv/bulk/sst_batcher.go | `sz` | |
| -pkg/kv/bulk/sst_batcher.go | `timing` | |
| pkg/kv/kvserver/closedts/ctpb/service.go | `LAI` | |
| pkg/kv/kvserver/closedts/ctpb/service.go | `SeqNum` | |
| -pkg/kv/kvserver/concurrency/lock/locking.go | `Durability` | |
| -pkg/kv/kvserver/concurrency/lock/locking.go | `Strength` | |
| pkg/kv/kvserver/concurrency/lock/locking.go | `WaitPolicy` | |
| pkg/kv/kvserver/kvserverpb/raft.go | `SnapshotRequest_Type` | |
| pkg/roachpb/data.go | `LeaseSequence` | |
| diff --git a/docs/generated/settings/settings-for-tenants.txt b/docs/generated/settings/settings-for-tenants.txt | |
| index 5c71d3dd71..655feef2c0 100644 | |
| --- a/docs/generated/settings/settings-for-tenants.txt | |
| +++ b/docs/generated/settings/settings-for-tenants.txt | |
| @@ -65,7 +65,7 @@ server.oidc_authentication.provider_url string sets OIDC provider URL ({provide | |
| server.oidc_authentication.redirect_url string https://localhost:8080/oidc/v1/callback sets OIDC redirect URL via a URL string or a JSON string containing a required `redirect_urls` key with an object that maps from region keys to URL strings (URLs should point to your load balancer and must route to the path /oidc/v1/callback) | |
| server.oidc_authentication.scopes string openid sets OIDC scopes to include with authentication request (space delimited list of strings, required to start with `openid`) | |
| server.rangelog.ttl duration 720h0m0s if nonzero, range log entries older than this duration are deleted every 10m0s. Should not be lowered below 24 hours. | |
| -server.shutdown.connection_wait duration 0s the maximum amount of time a server waits for all SQL connections to be closed before proceeding with a drain. When all SQL connections are closed before times out, the server early exits and proceeds to draining range leases. (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting) | |
| +server.shutdown.connection_wait duration 0s the maximum amount of time a server waits for all SQL connections to be closed before proceeding with a drain. (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting) | |
| server.shutdown.drain_wait duration 0s the amount of time a server waits in an unready state before proceeding with a drain (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting. --drain-wait is to specify the duration of the whole draining process, while server.shutdown.drain_wait is to set the wait time for health probes to notice that the node is not ready.) | |
| server.shutdown.lease_transfer_wait duration 5s the timeout for a single iteration of the range lease transfer phase of draining (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting) | |
| server.shutdown.query_wait duration 10s the timeout for waiting for active queries to finish during a drain (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting) | |
| @@ -83,7 +83,6 @@ server.web_session.purge.max_deletions_per_cycle integer 10 the maximum number o | |
| server.web_session.purge.period duration 1h0m0s the time until old sessions are deleted | |
| server.web_session.purge.ttl duration 1h0m0s if nonzero, entries in system.web_sessions older than this duration are periodically purged | |
| server.web_session_timeout duration 168h0m0s the duration that a newly created web session will be valid | |
| -sql.auth.resolve_membership_single_scan.enabled boolean true determines whether to populate the role membership cache with a single scan | |
| sql.contention.event_store.capacity byte size 64 MiB the in-memory storage capacity per-node of contention event store | |
| sql.contention.txn_id_cache.max_size byte size 0 B the maximum byte size TxnID cache will use (set to 0 to disable) | |
| sql.cross_db_fks.enabled boolean false if true, creating foreign key references across databases is allowed | |
| @@ -186,6 +185,5 @@ timeseries.storage.resolution_30m.ttl duration 2160h0m0s the maximum age of time | |
| trace.debug.enable boolean false if set, traces for recent requests can be seen at https://<ui>/debug/requests | |
| trace.jaeger.agent string the address of a Jaeger agent to receive traces using the Jaeger UDP Thrift protocol, as <host>:<port>. If no port is specified, 6381 will be used. | |
| trace.opentelemetry.collector string address of an OpenTelemetry trace collector to receive traces using the otel gRPC protocol, as <host>:<port>. If no port is specified, 4317 will be used. | |
| -trace.span_registry.enabled boolean true if set, ongoing traces can be seen at https://<ui>/debug/tracez | |
| trace.zipkin.collector string the address of a Zipkin instance to receive traces, as <host>:<port>. If no port is specified, 9411 will be used. | |
| -version version 21.2-88 set the active cluster version in the format '<major>.<minor>' | |
| +version version 21.2-84 set the active cluster version in the format '<major>.<minor>' | |
| diff --git a/docs/generated/settings/settings.html b/docs/generated/settings/settings.html | |
| index 1a431a87aa..8ff0ff55ef 100644 | |
| --- a/docs/generated/settings/settings.html | |
| +++ b/docs/generated/settings/settings.html | |
| @@ -77,7 +77,7 @@ | |
| <tr><td><code>server.oidc_authentication.redirect_url</code></td><td>string</td><td><code>https://localhost:8080/oidc/v1/callback</code></td><td>sets OIDC redirect URL via a URL string or a JSON string containing a required `redirect_urls` key with an object that maps from region keys to URL strings (URLs should point to your load balancer and must route to the path /oidc/v1/callback) </td></tr> | |
| <tr><td><code>server.oidc_authentication.scopes</code></td><td>string</td><td><code>openid</code></td><td>sets OIDC scopes to include with authentication request (space delimited list of strings, required to start with `openid`)</td></tr> | |
| <tr><td><code>server.rangelog.ttl</code></td><td>duration</td><td><code>720h0m0s</code></td><td>if nonzero, range log entries older than this duration are deleted every 10m0s. Should not be lowered below 24 hours.</td></tr> | |
| -<tr><td><code>server.shutdown.connection_wait</code></td><td>duration</td><td><code>0s</code></td><td>the maximum amount of time a server waits for all SQL connections to be closed before proceeding with a drain. When all SQL connections are closed before times out, the server early exits and proceeds to draining range leases. (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting)</td></tr> | |
| +<tr><td><code>server.shutdown.connection_wait</code></td><td>duration</td><td><code>0s</code></td><td>the maximum amount of time a server waits for all SQL connections to be closed before proceeding with a drain. (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting)</td></tr> | |
| <tr><td><code>server.shutdown.drain_wait</code></td><td>duration</td><td><code>0s</code></td><td>the amount of time a server waits in an unready state before proceeding with a drain (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting. --drain-wait is to specify the duration of the whole draining process, while server.shutdown.drain_wait is to set the wait time for health probes to notice that the node is not ready.)</td></tr> | |
| <tr><td><code>server.shutdown.lease_transfer_wait</code></td><td>duration</td><td><code>5s</code></td><td>the timeout for a single iteration of the range lease transfer phase of draining (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting)</td></tr> | |
| <tr><td><code>server.shutdown.query_wait</code></td><td>duration</td><td><code>10s</code></td><td>the timeout for waiting for active queries to finish during a drain (note that the --drain-wait parameter for cockroach node drain may need adjustment after changing this setting)</td></tr> | |
| @@ -95,7 +95,6 @@ | |
| <tr><td><code>server.web_session.purge.period</code></td><td>duration</td><td><code>1h0m0s</code></td><td>the time until old sessions are deleted</td></tr> | |
| <tr><td><code>server.web_session.purge.ttl</code></td><td>duration</td><td><code>1h0m0s</code></td><td>if nonzero, entries in system.web_sessions older than this duration are periodically purged</td></tr> | |
| <tr><td><code>server.web_session_timeout</code></td><td>duration</td><td><code>168h0m0s</code></td><td>the duration that a newly created web session will be valid</td></tr> | |
| -<tr><td><code>sql.auth.resolve_membership_single_scan.enabled</code></td><td>boolean</td><td><code>true</code></td><td>determines whether to populate the role membership cache with a single scan</td></tr> | |
| <tr><td><code>sql.contention.event_store.capacity</code></td><td>byte size</td><td><code>64 MiB</code></td><td>the in-memory storage capacity per-node of contention event store</td></tr> | |
| <tr><td><code>sql.contention.txn_id_cache.max_size</code></td><td>byte size</td><td><code>0 B</code></td><td>the maximum byte size TxnID cache will use (set to 0 to disable)</td></tr> | |
| <tr><td><code>sql.cross_db_fks.enabled</code></td><td>boolean</td><td><code>false</code></td><td>if true, creating foreign key references across databases is allowed</td></tr> | |
| @@ -199,8 +198,7 @@ | |
| <tr><td><code>trace.debug.enable</code></td><td>boolean</td><td><code>false</code></td><td>if set, traces for recent requests can be seen at https://<ui>/debug/requests</td></tr> | |
| <tr><td><code>trace.jaeger.agent</code></td><td>string</td><td><code></code></td><td>the address of a Jaeger agent to receive traces using the Jaeger UDP Thrift protocol, as <host>:<port>. If no port is specified, 6381 will be used.</td></tr> | |
| <tr><td><code>trace.opentelemetry.collector</code></td><td>string</td><td><code></code></td><td>address of an OpenTelemetry trace collector to receive traces using the otel gRPC protocol, as <host>:<port>. If no port is specified, 4317 will be used.</td></tr> | |
| -<tr><td><code>trace.span_registry.enabled</code></td><td>boolean</td><td><code>true</code></td><td>if set, ongoing traces can be seen at https://<ui>/debug/tracez</td></tr> | |
| <tr><td><code>trace.zipkin.collector</code></td><td>string</td><td><code></code></td><td>the address of a Zipkin instance to receive traces, as <host>:<port>. If no port is specified, 9411 will be used.</td></tr> | |
| -<tr><td><code>version</code></td><td>version</td><td><code>21.2-88</code></td><td>set the active cluster version in the format '<major>.<minor>'</td></tr> | |
| +<tr><td><code>version</code></td><td>version</td><td><code>21.2-84</code></td><td>set the active cluster version in the format '<major>.<minor>'</td></tr> | |
| </tbody> | |
| </table> | |
| diff --git a/docs/generated/sql/bnf/BUILD.bazel b/docs/generated/sql/bnf/BUILD.bazel | |
| index 00537a4b1f..706e8a799f 100644 | |
| --- a/docs/generated/sql/bnf/BUILD.bazel | |
| +++ b/docs/generated/sql/bnf/BUILD.bazel | |
| @@ -38,7 +38,6 @@ FILES = [ | |
| "alter_table_partition_by", | |
| "alter_table_set_schema_stmt", | |
| "alter_table_stmt", | |
| - "alter_tenant_csetting_stmt", | |
| "alter_type", | |
| "alter_view", | |
| "alter_view_owner_stmt", | |
| @@ -169,7 +168,6 @@ FILES = [ | |
| "select_stmt", | |
| "set_cluster_setting", | |
| "set_csetting_stmt", | |
| - "set_or_reset_csetting_stmt", | |
| "set_exprs_internal", | |
| "set_local_stmt", | |
| "set_operation", | |
| @@ -192,7 +190,6 @@ FILES = [ | |
| "show_indexes_stmt", | |
| "show_jobs", | |
| "show_keys", | |
| - "show_local_or_tenant_csettings_stmt", | |
| "show_locality", | |
| "show_locality_stmt", | |
| "show_partitions_stmt", | |
| diff --git a/docs/generated/sql/bnf/alter_stmt.bnf b/docs/generated/sql/bnf/alter_stmt.bnf | |
| index c41f5092df..50297fc693 100644 | |
| --- a/docs/generated/sql/bnf/alter_stmt.bnf | |
| +++ b/docs/generated/sql/bnf/alter_stmt.bnf | |
| @@ -1,4 +1,3 @@ | |
| alter_stmt ::= | |
| alter_ddl_stmt | |
| | alter_role_stmt | |
| - | alter_tenant_csetting_stmt | |
| diff --git a/docs/generated/sql/bnf/alter_tenant_csetting_stmt.bnf b/docs/generated/sql/bnf/alter_tenant_csetting_stmt.bnf | |
| deleted file mode 100644 | |
| index efdb92f776..0000000000 | |
| --- a/docs/generated/sql/bnf/alter_tenant_csetting_stmt.bnf | |
| +++ /dev/null | |
| @@ -1,3 +0,0 @@ | |
| -alter_tenant_csetting_stmt ::= | |
| - 'ALTER' 'TENANT' d_expr set_or_reset_csetting_stmt | |
| - | 'ALTER' 'ALL' 'TENANTS' set_or_reset_csetting_stmt | |
| diff --git a/docs/generated/sql/bnf/set_or_reset_csetting_stmt.bnf b/docs/generated/sql/bnf/set_or_reset_csetting_stmt.bnf | |
| deleted file mode 100644 | |
| index 1155b92d4e..0000000000 | |
| --- a/docs/generated/sql/bnf/set_or_reset_csetting_stmt.bnf | |
| +++ /dev/null | |
| @@ -1,3 +0,0 @@ | |
| -set_or_reset_csetting_stmt ::= | |
| - reset_csetting_stmt | |
| - | set_csetting_stmt | |
| diff --git a/docs/generated/sql/bnf/show_local_or_tenant_csettings_stmt.bnf b/docs/generated/sql/bnf/show_local_or_tenant_csettings_stmt.bnf | |
| deleted file mode 100644 | |
| index 631a6d6375..0000000000 | |
| --- a/docs/generated/sql/bnf/show_local_or_tenant_csettings_stmt.bnf | |
| +++ /dev/null | |
| @@ -1,3 +0,0 @@ | |
| -show_local_or_tenant_csettings_stmt ::= | |
| - show_csettings_stmt | |
| - | show_csettings_stmt 'FOR' 'TENANT' d_expr | |
| diff --git a/docs/generated/sql/bnf/show_var.bnf b/docs/generated/sql/bnf/show_var.bnf | |
| index 62daa21435..ffb29fc3e8 100644 | |
| --- a/docs/generated/sql/bnf/show_var.bnf | |
| +++ b/docs/generated/sql/bnf/show_var.bnf | |
| @@ -4,7 +4,7 @@ show_stmt ::= | |
| | show_constraints_stmt | |
| | show_create_stmt | |
| | show_create_schedules_stmt | |
| - | show_local_or_tenant_csettings_stmt | |
| + | show_csettings_stmt | |
| | show_databases_stmt | |
| | show_enums_stmt | |
| | show_types_stmt | |
| diff --git a/docs/generated/sql/bnf/stmt_block.bnf b/docs/generated/sql/bnf/stmt_block.bnf | |
| index 5e4af31f1b..2141dc8aa5 100644 | |
| --- a/docs/generated/sql/bnf/stmt_block.bnf | |
| +++ b/docs/generated/sql/bnf/stmt_block.bnf | |
| @@ -135,7 +135,6 @@ fetch_cursor_stmt ::= | |
| alter_stmt ::= | |
| alter_ddl_stmt | |
| | alter_role_stmt | |
| - | alter_tenant_csetting_stmt | |
| backup_stmt ::= | |
| 'BACKUP' opt_backup_targets 'INTO' sconst_or_placeholder 'IN' string_or_placeholder_opt_list opt_as_of_clause opt_with_backup_options | |
| @@ -230,7 +229,7 @@ show_stmt ::= | |
| | show_constraints_stmt | |
| | show_create_stmt | |
| | show_create_schedules_stmt | |
| - | show_local_or_tenant_csettings_stmt | |
| + | show_csettings_stmt | |
| | show_databases_stmt | |
| | show_enums_stmt | |
| | show_types_stmt | |
| @@ -333,7 +332,6 @@ targets ::= | |
| | table_pattern ',' table_pattern_list | |
| | 'TABLE' table_pattern_list | |
| | 'TENANT' iconst64 | |
| - | 'TENANT' 'identifier' | |
| | 'DATABASE' name_list | |
| role_spec_list ::= | |
| @@ -458,10 +456,6 @@ alter_role_stmt ::= | |
| | 'ALTER' 'ROLE_ALL' 'ALL' opt_in_database set_or_reset_clause | |
| | 'ALTER' 'USER_ALL' 'ALL' opt_in_database set_or_reset_clause | |
| -alter_tenant_csetting_stmt ::= | |
| - 'ALTER' 'TENANT' d_expr set_or_reset_csetting_stmt | |
| - | 'ALTER' 'ALL' 'TENANTS' set_or_reset_csetting_stmt | |
| - | |
| opt_backup_targets ::= | |
| targets | |
| @@ -715,9 +709,12 @@ show_create_schedules_stmt ::= | |
| 'SHOW' 'CREATE' 'ALL' 'SCHEDULES' | |
| | 'SHOW' 'CREATE' 'SCHEDULE' a_expr | |
| -show_local_or_tenant_csettings_stmt ::= | |
| - show_csettings_stmt | |
| - | show_csettings_stmt 'FOR' 'TENANT' d_expr | |
| +show_csettings_stmt ::= | |
| + 'SHOW' 'CLUSTER' 'SETTING' var_name | |
| + | 'SHOW' 'CLUSTER' 'SETTING' 'ALL' | |
| + | 'SHOW' 'ALL' 'CLUSTER' 'SETTINGS' | |
| + | 'SHOW' 'CLUSTER' 'SETTINGS' | |
| + | 'SHOW' 'PUBLIC' 'CLUSTER' 'SETTINGS' | |
| show_databases_stmt ::= | |
| 'SHOW' 'DATABASES' with_comment | |
| @@ -1245,7 +1242,6 @@ unreserved_keyword ::= | |
| | 'TEMPLATE' | |
| | 'TEMPORARY' | |
| | 'TENANT' | |
| - | 'TENANTS' | |
| | 'TESTING_RELOCATE' | |
| | 'TEXT' | |
| | 'TIES' | |
| @@ -1505,35 +1501,6 @@ set_or_reset_clause ::= | |
| | 'RESET_ALL' 'ALL' | |
| | 'RESET' session_var | |
| -d_expr ::= | |
| - 'ICONST' | |
| - | 'FCONST' | |
| - | 'SCONST' | |
| - | 'BCONST' | |
| - | 'BITCONST' | |
| - | typed_literal | |
| - | interval_value | |
| - | 'TRUE' | |
| - | 'FALSE' | |
| - | 'NULL' | |
| - | column_path_with_star | |
| - | '@' iconst64 | |
| - | 'PLACEHOLDER' | |
| - | '(' a_expr ')' '.' '*' | |
| - | '(' a_expr ')' '.' unrestricted_name | |
| - | '(' a_expr ')' '.' '@' 'ICONST' | |
| - | '(' a_expr ')' | |
| - | func_expr | |
| - | select_with_parens | |
| - | labeled_row | |
| - | 'ARRAY' select_with_parens | |
| - | 'ARRAY' row | |
| - | 'ARRAY' array_expr | |
| - | |
| -set_or_reset_csetting_stmt ::= | |
| - reset_csetting_stmt | |
| - | set_csetting_stmt | |
| - | |
| as_of_clause ::= | |
| 'AS' 'OF' 'SYSTEM' 'TIME' a_expr | |
| @@ -1761,13 +1728,6 @@ with_comment ::= | |
| 'WITH' 'COMMENT' | |
| | | |
| -show_csettings_stmt ::= | |
| - 'SHOW' 'CLUSTER' 'SETTING' var_name | |
| - | 'SHOW' 'CLUSTER' 'SETTING' 'ALL' | |
| - | 'SHOW' 'ALL' 'CLUSTER' 'SETTINGS' | |
| - | 'SHOW' 'CLUSTER' 'SETTINGS' | |
| - | 'SHOW' 'PUBLIC' 'CLUSTER' 'SETTINGS' | |
| - | |
| opt_on_targets_roles ::= | |
| 'ON' targets_roles | |
| | | |
| @@ -2031,36 +1991,6 @@ alter_backup_cmds ::= | |
| role_options ::= | |
| ( role_option ) ( ( role_option ) )* | |
| -typed_literal ::= | |
| - func_name_no_crdb_extra 'SCONST' | |
| - | const_typename 'SCONST' | |
| - | |
| -interval_value ::= | |
| - 'INTERVAL' 'SCONST' opt_interval_qualifier | |
| - | 'INTERVAL' '(' iconst32 ')' 'SCONST' | |
| - | |
| -column_path_with_star ::= | |
| - column_path | |
| - | db_object_name_component '.' unrestricted_name '.' unrestricted_name '.' '*' | |
| - | db_object_name_component '.' unrestricted_name '.' '*' | |
| - | db_object_name_component '.' '*' | |
| - | |
| -func_expr ::= | |
| - func_application within_group_clause filter_clause over_clause | |
| - | func_expr_common_subexpr | |
| - | |
| -labeled_row ::= | |
| - row | |
| - | '(' row 'AS' name_list ')' | |
| - | |
| -row ::= | |
| - 'ROW' '(' opt_expr_list ')' | |
| - | expr_tuple_unambiguous | |
| - | |
| -array_expr ::= | |
| - '[' opt_expr_list ']' | |
| - | '[' array_expr_list ']' | |
| - | |
| backup_options ::= | |
| 'ENCRYPTION_PASSPHRASE' '=' string_or_placeholder | |
| | 'REVISION_HISTORY' | |
| @@ -2579,79 +2509,30 @@ role_option ::= | |
| | password_clause | |
| | valid_until_clause | |
| -func_name_no_crdb_extra ::= | |
| - type_function_name_no_crdb_extra | |
| - | prefixed_column_path | |
| - | |
| -const_typename ::= | |
| - numeric | |
| - | bit_without_length | |
| - | character_without_length | |
| - | const_datetime | |
| - | const_geo | |
| - | |
| -opt_interval_qualifier ::= | |
| - interval_qualifier | |
| - | | |
| - | |
| -iconst32 ::= | |
| +d_expr ::= | |
| 'ICONST' | |
| - | |
| -func_application ::= | |
| - func_name '(' ')' | |
| - | func_name '(' expr_list opt_sort_clause ')' | |
| - | func_name '(' 'ALL' expr_list opt_sort_clause ')' | |
| - | func_name '(' 'DISTINCT' expr_list ')' | |
| - | func_name '(' '*' ')' | |
| - | |
| -within_group_clause ::= | |
| - 'WITHIN' 'GROUP' '(' single_sort_clause ')' | |
| - | | |
| - | |
| -filter_clause ::= | |
| - 'FILTER' '(' 'WHERE' a_expr ')' | |
| - | | |
| - | |
| -over_clause ::= | |
| - 'OVER' window_specification | |
| - | 'OVER' window_name | |
| - | | |
| - | |
| -func_expr_common_subexpr ::= | |
| - 'COLLATION' 'FOR' '(' a_expr ')' | |
| - | 'CURRENT_DATE' | |
| - | 'CURRENT_SCHEMA' | |
| - | 'CURRENT_CATALOG' | |
| - | 'CURRENT_TIMESTAMP' | |
| - | 'CURRENT_TIME' | |
| - | 'LOCALTIMESTAMP' | |
| - | 'LOCALTIME' | |
| - | 'CURRENT_USER' | |
| - | 'CURRENT_ROLE' | |
| - | 'SESSION_USER' | |
| - | 'USER' | |
| - | 'CAST' '(' a_expr 'AS' cast_target ')' | |
| - | 'ANNOTATE_TYPE' '(' a_expr ',' typename ')' | |
| - | 'IF' '(' a_expr ',' a_expr ',' a_expr ')' | |
| - | 'IFERROR' '(' a_expr ',' a_expr ',' a_expr ')' | |
| - | 'IFERROR' '(' a_expr ',' a_expr ')' | |
| - | 'ISERROR' '(' a_expr ')' | |
| - | 'ISERROR' '(' a_expr ',' a_expr ')' | |
| - | 'NULLIF' '(' a_expr ',' a_expr ')' | |
| - | 'IFNULL' '(' a_expr ',' a_expr ')' | |
| - | 'COALESCE' '(' expr_list ')' | |
| - | special_function | |
| - | |
| -opt_expr_list ::= | |
| - expr_list | |
| - | | |
| - | |
| -expr_tuple_unambiguous ::= | |
| - '(' ')' | |
| - | '(' tuple1_unambiguous_values ')' | |
| - | |
| -array_expr_list ::= | |
| - ( array_expr ) ( ( ',' array_expr ) )* | |
| + | 'FCONST' | |
| + | 'SCONST' | |
| + | 'BCONST' | |
| + | 'BITCONST' | |
| + | typed_literal | |
| + | interval_value | |
| + | 'TRUE' | |
| + | 'FALSE' | |
| + | 'NULL' | |
| + | column_path_with_star | |
| + | '@' iconst64 | |
| + | 'PLACEHOLDER' | |
| + | '(' a_expr ')' '.' '*' | |
| + | '(' a_expr ')' '.' unrestricted_name | |
| + | '(' a_expr ')' '.' '@' 'ICONST' | |
| + | '(' a_expr ')' | |
| + | func_expr | |
| + | select_with_parens | |
| + | labeled_row | |
| + | 'ARRAY' select_with_parens | |
| + | 'ARRAY' row | |
| + | 'ARRAY' array_expr | |
| array_subscripts ::= | |
| ( array_subscript ) ( ( array_subscript ) )* | |
| @@ -2873,10 +2754,20 @@ type_func_name_no_crdb_extra_keyword ::= | |
| general_type_name ::= | |
| type_function_name_no_crdb_extra | |
| +iconst32 ::= | |
| + 'ICONST' | |
| + | |
| complex_type_name ::= | |
| general_type_name '.' unrestricted_name | |
| | general_type_name '.' unrestricted_name '.' unrestricted_name | |
| +const_typename ::= | |
| + numeric | |
| + | bit_without_length | |
| + | character_without_length | |
| + | const_datetime | |
| + | const_geo | |
| + | |
| bit_with_length ::= | |
| 'BIT' opt_varying '(' iconst32 ')' | |
| | 'VARBIT' '(' iconst32 ')' | |
| @@ -2954,109 +2845,35 @@ valid_until_clause ::= | |
| 'VALID' 'UNTIL' string_or_placeholder | |
| | 'VALID' 'UNTIL' 'NULL' | |
| -type_function_name_no_crdb_extra ::= | |
| - 'identifier' | |
| - | unreserved_keyword | |
| - | type_func_name_no_crdb_extra_keyword | |
| +typed_literal ::= | |
| + func_name_no_crdb_extra 'SCONST' | |
| + | const_typename 'SCONST' | |
| -numeric ::= | |
| - 'INT' | |
| - | 'INTEGER' | |
| - | 'SMALLINT' | |
| - | 'BIGINT' | |
| - | 'REAL' | |
| - | 'FLOAT' opt_float | |
| - | 'DOUBLE' 'PRECISION' | |
| - | 'DECIMAL' opt_numeric_modifiers | |
| - | 'DEC' opt_numeric_modifiers | |
| - | 'NUMERIC' opt_numeric_modifiers | |
| - | 'BOOLEAN' | |
| +interval_value ::= | |
| + 'INTERVAL' 'SCONST' opt_interval_qualifier | |
| + | 'INTERVAL' '(' iconst32 ')' 'SCONST' | |
| -bit_without_length ::= | |
| - 'BIT' | |
| - | 'BIT' 'VARYING' | |
| - | 'VARBIT' | |
| +column_path_with_star ::= | |
| + column_path | |
| + | db_object_name_component '.' unrestricted_name '.' unrestricted_name '.' '*' | |
| + | db_object_name_component '.' unrestricted_name '.' '*' | |
| + | db_object_name_component '.' '*' | |
| -character_without_length ::= | |
| - character_base | |
| +func_expr ::= | |
| + func_application within_group_clause filter_clause over_clause | |
| + | func_expr_common_subexpr | |
| -const_datetime ::= | |
| - 'DATE' | |
| - | 'TIME' opt_timezone | |
| - | 'TIME' '(' iconst32 ')' opt_timezone | |
| - | 'TIMETZ' | |
| - | 'TIMETZ' '(' iconst32 ')' | |
| - | 'TIMESTAMP' opt_timezone | |
| - | 'TIMESTAMP' '(' iconst32 ')' opt_timezone | |
| - | 'TIMESTAMPTZ' | |
| - | 'TIMESTAMPTZ' '(' iconst32 ')' | |
| +labeled_row ::= | |
| + row | |
| + | '(' row 'AS' name_list ')' | |
| -const_geo ::= | |
| - 'GEOGRAPHY' | |
| - | 'GEOMETRY' | |
| - | 'BOX2D' | |
| - | 'GEOMETRY' '(' geo_shape_type ')' | |
| - | 'GEOGRAPHY' '(' geo_shape_type ')' | |
| - | 'GEOMETRY' '(' geo_shape_type ',' signed_iconst ')' | |
| - | 'GEOGRAPHY' '(' geo_shape_type ',' signed_iconst ')' | |
| +row ::= | |
| + 'ROW' '(' opt_expr_list ')' | |
| + | expr_tuple_unambiguous | |
| -interval_qualifier ::= | |
| - 'YEAR' | |
| - | 'MONTH' | |
| - | 'DAY' | |
| - | 'HOUR' | |
| - | 'MINUTE' | |
| - | interval_second | |
| - | 'YEAR' 'TO' 'MONTH' | |
| - | 'DAY' 'TO' 'HOUR' | |
| - | 'DAY' 'TO' 'MINUTE' | |
| - | 'DAY' 'TO' interval_second | |
| - | 'HOUR' 'TO' 'MINUTE' | |
| - | 'HOUR' 'TO' interval_second | |
| - | 'MINUTE' 'TO' interval_second | |
| - | |
| -func_name ::= | |
| - type_function_name | |
| - | prefixed_column_path | |
| - | |
| -single_sort_clause ::= | |
| - 'ORDER' 'BY' sortby | |
| - | 'ORDER' 'BY' sortby ',' sortby_list | |
| - | |
| -window_specification ::= | |
| - '(' opt_existing_window_name opt_partition_clause opt_sort_clause opt_frame_clause ')' | |
| - | |
| -window_name ::= | |
| - name | |
| - | |
| -special_function ::= | |
| - 'CURRENT_DATE' '(' ')' | |
| - | 'CURRENT_SCHEMA' '(' ')' | |
| - | 'CURRENT_TIMESTAMP' '(' ')' | |
| - | 'CURRENT_TIMESTAMP' '(' a_expr ')' | |
| - | 'CURRENT_TIME' '(' ')' | |
| - | 'CURRENT_TIME' '(' a_expr ')' | |
| - | 'LOCALTIMESTAMP' '(' ')' | |
| - | 'LOCALTIMESTAMP' '(' a_expr ')' | |
| - | 'LOCALTIME' '(' ')' | |
| - | 'LOCALTIME' '(' a_expr ')' | |
| - | 'CURRENT_USER' '(' ')' | |
| - | 'SESSION_USER' '(' ')' | |
| - | 'EXTRACT' '(' extract_list ')' | |
| - | 'EXTRACT_DURATION' '(' extract_list ')' | |
| - | 'OVERLAY' '(' overlay_list ')' | |
| - | 'POSITION' '(' position_list ')' | |
| - | 'SUBSTRING' '(' substr_list ')' | |
| - | 'TRIM' '(' 'BOTH' trim_list ')' | |
| - | 'TRIM' '(' 'LEADING' trim_list ')' | |
| - | 'TRIM' '(' 'TRAILING' trim_list ')' | |
| - | 'TRIM' '(' trim_list ')' | |
| - | 'GREATEST' '(' expr_list ')' | |
| - | 'LEAST' '(' expr_list ')' | |
| - | |
| -tuple1_unambiguous_values ::= | |
| - a_expr ',' | |
| - | a_expr ',' expr_list | |
| +array_expr ::= | |
| + '[' opt_expr_list ']' | |
| + | '[' array_expr_list ']' | |
| array_subscript ::= | |
| '[' a_expr ']' | |
| @@ -3170,6 +2987,52 @@ join_qual ::= | |
| rowsfrom_list ::= | |
| ( rowsfrom_item ) ( ( ',' rowsfrom_item ) )* | |
| +type_function_name_no_crdb_extra ::= | |
| + 'identifier' | |
| + | unreserved_keyword | |
| + | type_func_name_no_crdb_extra_keyword | |
| + | |
| +numeric ::= | |
| + 'INT' | |
| + | 'INTEGER' | |
| + | 'SMALLINT' | |
| + | 'BIGINT' | |
| + | 'REAL' | |
| + | 'FLOAT' opt_float | |
| + | 'DOUBLE' 'PRECISION' | |
| + | 'DECIMAL' opt_numeric_modifiers | |
| + | 'DEC' opt_numeric_modifiers | |
| + | 'NUMERIC' opt_numeric_modifiers | |
| + | 'BOOLEAN' | |
| + | |
| +bit_without_length ::= | |
| + 'BIT' | |
| + | 'BIT' 'VARYING' | |
| + | 'VARBIT' | |
| + | |
| +character_without_length ::= | |
| + character_base | |
| + | |
| +const_datetime ::= | |
| + 'DATE' | |
| + | 'TIME' opt_timezone | |
| + | 'TIME' '(' iconst32 ')' opt_timezone | |
| + | 'TIMETZ' | |
| + | 'TIMETZ' '(' iconst32 ')' | |
| + | 'TIMESTAMP' opt_timezone | |
| + | 'TIMESTAMP' '(' iconst32 ')' opt_timezone | |
| + | 'TIMESTAMPTZ' | |
| + | 'TIMESTAMPTZ' '(' iconst32 ')' | |
| + | |
| +const_geo ::= | |
| + 'GEOGRAPHY' | |
| + | 'GEOMETRY' | |
| + | 'BOX2D' | |
| + | 'GEOMETRY' '(' geo_shape_type ')' | |
| + | 'GEOGRAPHY' '(' geo_shape_type ')' | |
| + | 'GEOMETRY' '(' geo_shape_type ',' signed_iconst ')' | |
| + | 'GEOGRAPHY' '(' geo_shape_type ',' signed_iconst ')' | |
| + | |
| opt_varying ::= | |
| 'VARYING' | |
| | | |
| @@ -3180,6 +3043,21 @@ character_base ::= | |
| | 'VARCHAR' | |
| | 'STRING' | |
| +interval_qualifier ::= | |
| + 'YEAR' | |
| + | 'MONTH' | |
| + | 'DAY' | |
| + | 'HOUR' | |
| + | 'MINUTE' | |
| + | interval_second | |
| + | 'YEAR' 'TO' 'MONTH' | |
| + | 'DAY' 'TO' 'HOUR' | |
| + | 'DAY' 'TO' 'MINUTE' | |
| + | 'DAY' 'TO' interval_second | |
| + | 'HOUR' 'TO' 'MINUTE' | |
| + | 'HOUR' 'TO' interval_second | |
| + | 'MINUTE' 'TO' interval_second | |
| + | |
| opt_column ::= | |
| 'COLUMN' | |
| | | |
| @@ -3235,101 +3113,69 @@ storage_parameter_key_list ::= | |
| partition_by_index ::= | |
| partition_by | |
| -opt_float ::= | |
| - '(' 'ICONST' ')' | |
| - | | |
| - | |
| -opt_numeric_modifiers ::= | |
| - '(' iconst32 ')' | |
| - | '(' iconst32 ',' iconst32 ')' | |
| - | | |
| +func_name_no_crdb_extra ::= | |
| + type_function_name_no_crdb_extra | |
| + | prefixed_column_path | |
| -opt_timezone ::= | |
| - 'WITH' 'TIME' 'ZONE' | |
| - | 'WITHOUT' 'TIME' 'ZONE' | |
| +opt_interval_qualifier ::= | |
| + interval_qualifier | |
| | | |
| -geo_shape_type ::= | |
| - 'POINT' | |
| - | 'POINTM' | |
| - | 'POINTZ' | |
| - | 'POINTZM' | |
| - | 'LINESTRING' | |
| - | 'LINESTRINGM' | |
| - | 'LINESTRINGZ' | |
| - | 'LINESTRINGZM' | |
| - | 'POLYGON' | |
| - | 'POLYGONM' | |
| - | 'POLYGONZ' | |
| - | 'POLYGONZM' | |
| - | 'MULTIPOINT' | |
| - | 'MULTIPOINTM' | |
| - | 'MULTIPOINTZ' | |
| - | 'MULTIPOINTZM' | |
| - | 'MULTILINESTRING' | |
| - | 'MULTILINESTRINGM' | |
| - | 'MULTILINESTRINGZ' | |
| - | 'MULTILINESTRINGZM' | |
| - | 'MULTIPOLYGON' | |
| - | 'MULTIPOLYGONM' | |
| - | 'MULTIPOLYGONZ' | |
| - | 'MULTIPOLYGONZM' | |
| - | 'GEOMETRYCOLLECTION' | |
| - | 'GEOMETRYCOLLECTIONM' | |
| - | 'GEOMETRYCOLLECTIONZ' | |
| - | 'GEOMETRYCOLLECTIONZM' | |
| - | 'GEOMETRY' | |
| - | 'GEOMETRYM' | |
| - | 'GEOMETRYZ' | |
| - | 'GEOMETRYZM' | |
| - | |
| -interval_second ::= | |
| - 'SECOND' | |
| - | 'SECOND' '(' iconst32 ')' | |
| - | |
| -type_function_name ::= | |
| - 'identifier' | |
| - | unreserved_keyword | |
| - | type_func_name_keyword | |
| +func_application ::= | |
| + func_name '(' ')' | |
| + | func_name '(' expr_list opt_sort_clause ')' | |
| + | func_name '(' 'ALL' expr_list opt_sort_clause ')' | |
| + | func_name '(' 'DISTINCT' expr_list ')' | |
| + | func_name '(' '*' ')' | |
| -opt_existing_window_name ::= | |
| - name | |
| +within_group_clause ::= | |
| + 'WITHIN' 'GROUP' '(' single_sort_clause ')' | |
| | | |
| -opt_partition_clause ::= | |
| - 'PARTITION' 'BY' expr_list | |
| +filter_clause ::= | |
| + 'FILTER' '(' 'WHERE' a_expr ')' | |
| | | |
| -opt_frame_clause ::= | |
| - 'RANGE' frame_extent opt_frame_exclusion | |
| - | 'ROWS' frame_extent opt_frame_exclusion | |
| - | 'GROUPS' frame_extent opt_frame_exclusion | |
| +over_clause ::= | |
| + 'OVER' window_specification | |
| + | 'OVER' window_name | |
| | | |
| -extract_list ::= | |
| - extract_arg 'FROM' a_expr | |
| - | expr_list | |
| - | |
| -overlay_list ::= | |
| - a_expr overlay_placing substr_from substr_for | |
| - | a_expr overlay_placing substr_from | |
| - | expr_list | |
| +func_expr_common_subexpr ::= | |
| + 'COLLATION' 'FOR' '(' a_expr ')' | |
| + | 'CURRENT_DATE' | |
| + | 'CURRENT_SCHEMA' | |
| + | 'CURRENT_CATALOG' | |
| + | 'CURRENT_TIMESTAMP' | |
| + | 'CURRENT_TIME' | |
| + | 'LOCALTIMESTAMP' | |
| + | 'LOCALTIME' | |
| + | 'CURRENT_USER' | |
| + | 'CURRENT_ROLE' | |
| + | 'SESSION_USER' | |
| + | 'USER' | |
| + | 'CAST' '(' a_expr 'AS' cast_target ')' | |
| + | 'ANNOTATE_TYPE' '(' a_expr ',' typename ')' | |
| + | 'IF' '(' a_expr ',' a_expr ',' a_expr ')' | |
| + | 'IFERROR' '(' a_expr ',' a_expr ',' a_expr ')' | |
| + | 'IFERROR' '(' a_expr ',' a_expr ')' | |
| + | 'ISERROR' '(' a_expr ')' | |
| + | 'ISERROR' '(' a_expr ',' a_expr ')' | |
| + | 'NULLIF' '(' a_expr ',' a_expr ')' | |
| + | 'IFNULL' '(' a_expr ',' a_expr ')' | |
| + | 'COALESCE' '(' expr_list ')' | |
| + | special_function | |
| -position_list ::= | |
| - b_expr 'IN' b_expr | |
| +opt_expr_list ::= | |
| + expr_list | |
| | | |
| -substr_list ::= | |
| - a_expr substr_from substr_for | |
| - | a_expr substr_for substr_from | |
| - | a_expr substr_from | |
| - | a_expr substr_for | |
| - | opt_expr_list | |
| +expr_tuple_unambiguous ::= | |
| + '(' ')' | |
| + | '(' tuple1_unambiguous_values ')' | |
| -trim_list ::= | |
| - a_expr 'FROM' expr_list | |
| - | 'FROM' expr_list | |
| - | expr_list | |
| +array_expr_list ::= | |
| + ( array_expr ) ( ( ',' array_expr ) )* | |
| opt_slice_bound ::= | |
| a_expr | |
| @@ -3383,10 +3229,62 @@ join_outer ::= | |
| rowsfrom_item ::= | |
| func_expr_windowless | |
| +opt_float ::= | |
| + '(' 'ICONST' ')' | |
| + | | |
| + | |
| +opt_numeric_modifiers ::= | |
| + '(' iconst32 ')' | |
| + | '(' iconst32 ',' iconst32 ')' | |
| + | | |
| + | |
| +opt_timezone ::= | |
| + 'WITH' 'TIME' 'ZONE' | |
| + | 'WITHOUT' 'TIME' 'ZONE' | |
| + | | |
| + | |
| +geo_shape_type ::= | |
| + 'POINT' | |
| + | 'POINTM' | |
| + | 'POINTZ' | |
| + | 'POINTZM' | |
| + | 'LINESTRING' | |
| + | 'LINESTRINGM' | |
| + | 'LINESTRINGZ' | |
| + | 'LINESTRINGZM' | |
| + | 'POLYGON' | |
| + | 'POLYGONM' | |
| + | 'POLYGONZ' | |
| + | 'POLYGONZM' | |
| + | 'MULTIPOINT' | |
| + | 'MULTIPOINTM' | |
| + | 'MULTIPOINTZ' | |
| + | 'MULTIPOINTZM' | |
| + | 'MULTILINESTRING' | |
| + | 'MULTILINESTRINGM' | |
| + | 'MULTILINESTRINGZ' | |
| + | 'MULTILINESTRINGZM' | |
| + | 'MULTIPOLYGON' | |
| + | 'MULTIPOLYGONM' | |
| + | 'MULTIPOLYGONZ' | |
| + | 'MULTIPOLYGONZM' | |
| + | 'GEOMETRYCOLLECTION' | |
| + | 'GEOMETRYCOLLECTIONM' | |
| + | 'GEOMETRYCOLLECTIONZ' | |
| + | 'GEOMETRYCOLLECTIONZM' | |
| + | 'GEOMETRY' | |
| + | 'GEOMETRYM' | |
| + | 'GEOMETRYZ' | |
| + | 'GEOMETRYZM' | |
| + | |
| char_aliases ::= | |
| 'CHAR' | |
| | 'CHARACTER' | |
| +interval_second ::= | |
| + 'SECOND' | |
| + | 'SECOND' '(' iconst32 ')' | |
| + | |
| col_qual_list ::= | |
| ( ) ( ( col_qualification ) )* | |
| @@ -3402,35 +3300,48 @@ reference_actions ::= | |
| | reference_on_delete reference_on_update | |
| | | |
| -frame_extent ::= | |
| - frame_bound | |
| - | 'BETWEEN' frame_bound 'AND' frame_bound | |
| +func_name ::= | |
| + type_function_name | |
| + | prefixed_column_path | |
| -opt_frame_exclusion ::= | |
| - 'EXCLUDE' 'CURRENT' 'ROW' | |
| - | 'EXCLUDE' 'GROUP' | |
| - | 'EXCLUDE' 'TIES' | |
| - | 'EXCLUDE' 'NO' 'OTHERS' | |
| - | | |
| +single_sort_clause ::= | |
| + 'ORDER' 'BY' sortby | |
| + | 'ORDER' 'BY' sortby ',' sortby_list | |
| -extract_arg ::= | |
| - 'identifier' | |
| - | 'YEAR' | |
| - | 'MONTH' | |
| - | 'DAY' | |
| - | 'HOUR' | |
| - | 'MINUTE' | |
| - | 'SECOND' | |
| - | 'SCONST' | |
| +window_specification ::= | |
| + '(' opt_existing_window_name opt_partition_clause opt_sort_clause opt_frame_clause ')' | |
| -overlay_placing ::= | |
| - 'PLACING' a_expr | |
| +window_name ::= | |
| + name | |
| -substr_from ::= | |
| - 'FROM' a_expr | |
| +special_function ::= | |
| + 'CURRENT_DATE' '(' ')' | |
| + | 'CURRENT_SCHEMA' '(' ')' | |
| + | 'CURRENT_TIMESTAMP' '(' ')' | |
| + | 'CURRENT_TIMESTAMP' '(' a_expr ')' | |
| + | 'CURRENT_TIME' '(' ')' | |
| + | 'CURRENT_TIME' '(' a_expr ')' | |
| + | 'LOCALTIMESTAMP' '(' ')' | |
| + | 'LOCALTIMESTAMP' '(' a_expr ')' | |
| + | 'LOCALTIME' '(' ')' | |
| + | 'LOCALTIME' '(' a_expr ')' | |
| + | 'CURRENT_USER' '(' ')' | |
| + | 'SESSION_USER' '(' ')' | |
| + | 'EXTRACT' '(' extract_list ')' | |
| + | 'EXTRACT_DURATION' '(' extract_list ')' | |
| + | 'OVERLAY' '(' overlay_list ')' | |
| + | 'POSITION' '(' position_list ')' | |
| + | 'SUBSTRING' '(' substr_list ')' | |
| + | 'TRIM' '(' 'BOTH' trim_list ')' | |
| + | 'TRIM' '(' 'LEADING' trim_list ')' | |
| + | 'TRIM' '(' 'TRAILING' trim_list ')' | |
| + | 'TRIM' '(' trim_list ')' | |
| + | 'GREATEST' '(' expr_list ')' | |
| + | 'LEAST' '(' expr_list ')' | |
| -substr_for ::= | |
| - 'FOR' a_expr | |
| +tuple1_unambiguous_values ::= | |
| + a_expr ',' | |
| + | a_expr ',' expr_list | |
| list_partition ::= | |
| partition 'VALUES' 'IN' '(' expr_list ')' opt_partition_by | |
| @@ -3469,12 +3380,49 @@ reference_on_update ::= | |
| reference_on_delete ::= | |
| 'ON' 'DELETE' reference_action | |
| -frame_bound ::= | |
| - 'UNBOUNDED' 'PRECEDING' | |
| - | 'UNBOUNDED' 'FOLLOWING' | |
| - | 'CURRENT' 'ROW' | |
| - | a_expr 'PRECEDING' | |
| - | a_expr 'FOLLOWING' | |
| +type_function_name ::= | |
| + 'identifier' | |
| + | unreserved_keyword | |
| + | type_func_name_keyword | |
| + | |
| +opt_existing_window_name ::= | |
| + name | |
| + | | |
| + | |
| +opt_partition_clause ::= | |
| + 'PARTITION' 'BY' expr_list | |
| + | | |
| + | |
| +opt_frame_clause ::= | |
| + 'RANGE' frame_extent opt_frame_exclusion | |
| + | 'ROWS' frame_extent opt_frame_exclusion | |
| + | 'GROUPS' frame_extent opt_frame_exclusion | |
| + | | |
| + | |
| +extract_list ::= | |
| + extract_arg 'FROM' a_expr | |
| + | expr_list | |
| + | |
| +overlay_list ::= | |
| + a_expr overlay_placing substr_from substr_for | |
| + | a_expr overlay_placing substr_from | |
| + | expr_list | |
| + | |
| +position_list ::= | |
| + b_expr 'IN' b_expr | |
| + | | |
| + | |
| +substr_list ::= | |
| + a_expr substr_from substr_for | |
| + | a_expr substr_for substr_from | |
| + | a_expr substr_from | |
| + | a_expr substr_for | |
| + | opt_expr_list | |
| + | |
| +trim_list ::= | |
| + a_expr 'FROM' expr_list | |
| + | 'FROM' expr_list | |
| + | expr_list | |
| opt_partition_by ::= | |
| partition_by | |
| @@ -3508,6 +3456,36 @@ reference_action ::= | |
| | 'SET' 'NULL' | |
| | 'SET' 'DEFAULT' | |
| +frame_extent ::= | |
| + frame_bound | |
| + | 'BETWEEN' frame_bound 'AND' frame_bound | |
| + | |
| +opt_frame_exclusion ::= | |
| + 'EXCLUDE' 'CURRENT' 'ROW' | |
| + | 'EXCLUDE' 'GROUP' | |
| + | 'EXCLUDE' 'TIES' | |
| + | 'EXCLUDE' 'NO' 'OTHERS' | |
| + | | |
| + | |
| +extract_arg ::= | |
| + 'identifier' | |
| + | 'YEAR' | |
| + | 'MONTH' | |
| + | 'DAY' | |
| + | 'HOUR' | |
| + | 'MINUTE' | |
| + | 'SECOND' | |
| + | 'SCONST' | |
| + | |
| +overlay_placing ::= | |
| + 'PLACING' a_expr | |
| + | |
| +substr_from ::= | |
| + 'FROM' a_expr | |
| + | |
| +substr_for ::= | |
| + 'FOR' a_expr | |
| + | |
| opt_name_parens ::= | |
| '(' name ')' | |
| | | |
| @@ -3521,3 +3499,10 @@ generated_always_as ::= | |
| generated_by_default_as ::= | |
| 'GENERATED_BY_DEFAULT' 'BY' 'DEFAULT' 'AS' | |
| + | |
| +frame_bound ::= | |
| + 'UNBOUNDED' 'PRECEDING' | |
| + | 'UNBOUNDED' 'FOLLOWING' | |
| + | 'CURRENT' 'ROW' | |
| + | a_expr 'PRECEDING' | |
| + | a_expr 'FOLLOWING' | |
| diff --git a/docs/generated/sql/functions.md b/docs/generated/sql/functions.md | |
| index d3c38de22b..2038a9972f 100644 | |
| --- a/docs/generated/sql/functions.md | |
| +++ b/docs/generated/sql/functions.md | |
| @@ -2910,8 +2910,6 @@ a CockroachDB HLC in decimal form.</p> | |
| <p>Note that uses of this function disable server-side optimizations and | |
| may increase either contention or retry errors, or both.</p> | |
| </span></td></tr> | |
| -<tr><td><a name="crdb_internal.active_version"></a><code>crdb_internal.active_version() → jsonb</code></td><td><span class="funcdesc"><p>Returns the current active cluster version.</p> | |
| -</span></td></tr> | |
| <tr><td><a name="crdb_internal.approximate_timestamp"></a><code>crdb_internal.approximate_timestamp(timestamp: <a href="decimal.html">decimal</a>) → <a href="timestamp.html">timestamp</a></code></td><td><span class="funcdesc"><p>Converts the crdb_internal_mvcc_timestamp column into an approximate timestamp.</p> | |
| </span></td></tr> | |
| <tr><td><a name="crdb_internal.assignment_cast"></a><code>crdb_internal.assignment_cast(val: anyelement, type: anyelement) → anyelement</code></td><td><span class="funcdesc"><p>This function is used internally to perform assignment casts during mutations.</p> | |
| diff --git a/docs/generated/swagger/spec.json b/docs/generated/swagger/spec.json | |
| index 6e5daf8aa9..5047ea7fe1 100644 | |
| --- a/docs/generated/swagger/spec.json | |
| +++ b/docs/generated/swagger/spec.json | |
| @@ -1023,12 +1023,6 @@ | |
| }, | |
| "x-go-package": "github.com/cockroachdb/cockroach/pkg/util/metric" | |
| }, | |
| - "RangeID": { | |
| - "type": "integer", | |
| - "format": "int64", | |
| - "title": "A RangeID is a unique ID associated to a Raft consensus group.", | |
| - "x-go-package": "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - }, | |
| "RangeProblems": { | |
| "type": "object", | |
| "title": "RangeProblems describes issues reported by a range. For internal use only.", | |
| @@ -1580,12 +1574,12 @@ | |
| "format": "int32", | |
| "x-go-name": "Internal" | |
| }, | |
| - "major": { | |
| + "major_val": { | |
| "type": "integer", | |
| "format": "int32", | |
| "x-go-name": "Major" | |
| }, | |
| - "minor": { | |
| + "minor_val": { | |
| "type": "integer", | |
| "format": "int32", | |
| "x-go-name": "Minor" | |
| @@ -1794,54 +1788,6 @@ | |
| }, | |
| "x-go-package": "github.com/cockroachdb/cockroach/pkg/server" | |
| }, | |
| - "hotRangeInfo": { | |
| - "description": "(ie its range ID, QPS, table name, etc.).", | |
| - "type": "object", | |
| - "title": "Hot range details struct describes common information about hot range,", | |
| - "properties": { | |
| - "database_name": { | |
| - "type": "string", | |
| - "x-go-name": "DatabaseName" | |
| - }, | |
| - "index_name": { | |
| - "type": "string", | |
| - "x-go-name": "IndexName" | |
| - }, | |
| - "leaseholder_node_id": { | |
| - "$ref": "#/definitions/NodeID" | |
| - }, | |
| - "node_id": { | |
| - "$ref": "#/definitions/NodeID" | |
| - }, | |
| - "qps": { | |
| - "type": "number", | |
| - "format": "double", | |
| - "x-go-name": "QPS" | |
| - }, | |
| - "range_id": { | |
| - "$ref": "#/definitions/RangeID" | |
| - }, | |
| - "replica_node_ids": { | |
| - "type": "array", | |
| - "items": { | |
| - "$ref": "#/definitions/NodeID" | |
| - }, | |
| - "x-go-name": "ReplicaNodeIDs" | |
| - }, | |
| - "schema_name": { | |
| - "type": "string", | |
| - "x-go-name": "SchemaName" | |
| - }, | |
| - "store_id": { | |
| - "$ref": "#/definitions/StoreID" | |
| - }, | |
| - "table_name": { | |
| - "type": "string", | |
| - "x-go-name": "TableName" | |
| - } | |
| - }, | |
| - "x-go-package": "github.com/cockroachdb/cockroach/pkg/server" | |
| - }, | |
| "hotRangesResponse": { | |
| "type": "object", | |
| "title": "Response struct for listHotRanges.", | |
| @@ -1851,12 +1797,15 @@ | |
| "type": "string", | |
| "x-go-name": "Next" | |
| }, | |
| - "ranges": { | |
| - "type": "array", | |
| - "items": { | |
| - "$ref": "#/definitions/hotRangeInfo" | |
| + "ranges_by_node_id": { | |
| + "type": "object", | |
| + "additionalProperties": { | |
| + "type": "array", | |
| + "items": { | |
| + "$ref": "#/definitions/rangeDescriptorInfo" | |
| + } | |
| }, | |
| - "x-go-name": "Ranges" | |
| + "x-go-name": "RangesByNodeID" | |
| }, | |
| "response_error": { | |
| "type": "array", | |
| diff --git a/docs/tech-notes/change-replicas.md b/docs/tech-notes/change-replicas.md | |
| deleted file mode 100644 | |
| index 5de72a42db..0000000000 | |
| --- a/docs/tech-notes/change-replicas.md | |
| +++ /dev/null | |
| @@ -1,93 +0,0 @@ | |
| -# Up-replicating replicas with snapshots | |
| - | |
| -This tech note briefly explains the end-to-end process of up-replicating a range, starting from the `replicateQueue` to | |
| -sending Learner snapshots. This is meant to serve as a high level overview of the code paths taken, and more detailed | |
| -descriptions can be found in tech notes specific to each area. | |
| - | |
| -## Introduction | |
| - | |
| -Raft snapshots are necessary when a follower replica is unable to catch up using the existing Raft logs of the leader. | |
| -Such a scenario occurs when an existing replica falls too far behind the rest of the Raft group such that we've | |
| -truncated the Raft log above it. On the other hand, Learner snapshots are necessary when we add a new replica (due to a | |
| -rebalancing operation) and the new replica needs a snapshot to up-replicate. We should make a small distinction | |
| -that `Learner snapshots` are sent during rebalancing operations by the `replicateQueue` and `Raft snapshots` | |
| -are sent by the `raftSnapshotQueue` when a replica falls behind. However, there is no difference in the way the | |
| -snapshots are sent, generated or applied. For example, in the case where the replication factor is increased, the Raft | |
| -group would need to create a new replica from a clean slate. To up-replicate this replica, we would send it a Learner | |
| -snapshot of the full range data. | |
| - | |
| -In this note, we will focus on the scenario where a Learner snapshot is needed for the up-replication and rebalancing of | |
| -replicas. | |
| - | |
| -## ReplicateQueue | |
| - | |
| -As a brief overview, the `replicateQueue` manages a queue of replicas and is responsible for processing replicas for | |
| -rebalancing. For each replica that holds the lease, it invokes the allocator to compute decisions on whether any placement changes are needed | |
| -and the `replicateQueue` executes these changes. Then the `replicateQueue` calls `changeReplicas` to act on these change | |
| -requests and repeats until all changes are complete. | |
| - | |
| -## ChangeReplicas API | |
| - | |
| -The `AdminChangeReplicas` API exposed by KV is mainly responsible for atomically changing replicas of a range. These | |
| -changes include non-voter promotions, voter/non-voter swaps, additions and removals. The change is performed in a | |
| -distributed transaction and takes effect when the transaction is committed. There are many details involved in these | |
| -changes, but for the purposes of this note, we will focus on up-replication and the process of sending Learner | |
| -snapshots. | |
| - | |
| -During up-replication, `ChangeReplicas` runs a transaction to add a new learner replica to the range. Learner | |
| -replicas are new replicas that receive Raft traffic but do not participate in quorum; allowing the range to remain | |
| -highly available during the replication process. | |
| - | |
| -Once the learners have been added to the range, they are synchronously sent a Learner snapshot from the leaseholder | |
| -(which is typically, but not necessarily, the Raft leader) to up-replicate. There is some nuance here as Raft snapshots | |
| -are typically automatically sent by the existing Raft snapshot queue. However, we are synchronously sending a snapshot | |
| -here in the replicateQueue to quickly catch up learners. To prevent a race with the raftSnapshotQueue, we lock snapshots | |
| -to learners and non-voters on the current leaseholder store while processing the change. In addition, we also place a | |
| -lock on log truncation to ensure we don't truncate the Raft | |
| -log while a snapshot is inflight, preventing wasted snapshots. However, both of these locks are the best effort as | |
| -opposed to guaranteed as the lease can be transferred and the new leaseholder could still truncate the log. | |
| - | |
| -## Sending Snapshots | |
| - | |
| -The snapshot process itself is broken into three parts: generating, transmitting and applying the snapshot. This process | |
| -is the same whether it is invoked by the `replicateQueue` or `raftSnapshotQueue`. | |
| - | |
| -### Generating the snapshot | |
| - | |
| -A snapshot is a bulk transfer of all replicated data in a range and everything the replica needs to be a member of a | |
| -Raft group. It consists of a consistent view of the state of some replica of a range as of an applied index. The | |
| -`GetSnapshot` method gets a storage engine snapshot which reflects the replicated state as of the applied index the | |
| -snapshot is generated at, and creates an iterator from the storage snapshot. A storage engine snapshot is important to | |
| -ensure that multiple iterators created at different points in time see a consistent replicated state that does not | |
| -change while the snapshot is streamed. In our current code, the engine snapshot is not necessary for correctness as only | |
| -one iterator is constructed. | |
| - | |
| -### Transmitting the snapshot | |
| - | |
| -The snapshot transfer is sent through a bi-directional stream of snapshot requests and snapshot responses. However, | |
| -before the actual range data is sent, the streaming rpc first sends a header message to the recipient store and blocks | |
| -until the store responds to accept or reject the snapshot data. | |
| - | |
| -The recipient checks the following conditions before accepting the snapshot. We currently allow one concurrent snapshot | |
| -application on a receiver store at a time. Consequently, if there are snapshot applications already in-flight on the | |
| -receiver, incoming snapshots are throttled or rejected if they wait too long. The receiver then checks whether its store | |
| -has a compatible replica present to have the snapshot applied. In the case of up-replication, we expect the learner | |
| -replica to be present on the receiver store. However, if the snapshot overlaps an existing replica or replica | |
| -placeholder on the receiver, the snapshot will be rejected as well. Once these conditions have been verified, a response | |
| -message will be sent back to the sender. | |
| - | |
| -Once the recipient has accepted, the sender proceeds to use the iterator to read chunks of | |
| -size `kv.snapshot_sender.batch_size` from storage into memory. The batch size is enforced as to not hold on to a | |
| -significant chunk of data in memory. Then, in-memory batches are created to send the snapshot data in streaming grpc | |
| -message chunks to the recipient. Note that snapshots are rate limited depending on cluster settings as to not overload | |
| -the network. Finally, once all the data is sent, the sender sends a final message to the receiver and waits for a | |
| -response from the recipient indicating whether the snapshot was a success. On the receiver side, the key-value pairs are | |
| -used to construct multiple SSTs for direct ingestion, which prevents the receiver from holding the entire snapshot in | |
| -memory. | |
| - | |
| -### Applying the snapshot | |
| - | |
| -Once the snapshot is received on the receiver, defensive checks are applied to ensure the correct snapshot is received. | |
| -It ensures there is an initialized learner replica or creates a placeholder to accept the snapshot. Finally, the Raft | |
| -snapshot message is handed to the replica’s Raft node, which will apply the snapshot. The placeholder or learner is then | |
| -converted to a fully initialized replica. | |
| diff --git a/docs/tech-notes/contexts.md b/docs/tech-notes/contexts.md | |
| index d9f8a1cc73..3eb14c671e 100644 | |
| --- a/docs/tech-notes/contexts.md | |
| +++ b/docs/tech-notes/contexts.md | |
| @@ -278,19 +278,8 @@ Of note: | |
| - a cancellable sub-context of a cancellable context means there is a | |
| goroutine spawn under you for propagating the parents cancellation, | |
| so it can be expensive. | |
| -- GRPC propagates cancel and deadline contexts across the wire. Deadline | |
| - contexts are just copied and recreated on other side. The way cancelation | |
| - works (for instance when a SQL statement timeout expires) is that after | |
| - sending a message GRPC does a | |
| - [select](https://github.com/cockroachdb/vendored/blob/ed53f102e86b51c228b61d3ad54f1428982fe75e/google.golang.org/grpc/internal/transport/transport.go#L178) | |
| - on the response and also listens on the Done channel for the context passed it | |
| - by the client. This in turn will close the GRPC stream which will send an | |
| - [RST](https://datatracker.ietf.org/doc/html/rfc7540#section-6.4) | |
| - message to the server which will cause the server side to be | |
| - [canceled](https://github.com/cockroachdb/vendored/blob/ed53f102e86b51c228b61d3ad54f1428982fe75e/google.golang.org/grpc/internal/transport/http2_server.go#L1231). | |
| - See also TestGRPCDeadlinePropagation and | |
| - TestTenantStatementTimeoutAdmissionQueueCancelation for tests that rely on this | |
| - behavior. | |
| +- it's stil la bit unclear how context cancellation crosses gRPC and | |
| + libpq boundaries. People are working on this. | |
| ## Technical notes | |
| diff --git a/go.mod b/go.mod | |
| index 8164331538..fadc980c32 100644 | |
| --- a/go.mod | |
| +++ b/go.mod | |
| @@ -46,7 +46,7 @@ require ( | |
| github.com/cockroachdb/go-test-teamcity v0.0.0-20191211140407-cff980ad0a55 | |
| github.com/cockroachdb/gostdlib v1.13.0 | |
| github.com/cockroachdb/logtags v0.0.0-20211118104740-dabe8e521a4f | |
| - github.com/cockroachdb/pebble v0.0.0-20220307192532-e2b7bb844759 | |
| + github.com/cockroachdb/pebble v0.0.0-20220227235451-40d39da505a5 | |
| github.com/cockroachdb/redact v1.1.3 | |
| github.com/cockroachdb/returncheck v0.0.0-20200612231554-92cdbca611dd | |
| github.com/cockroachdb/stress v0.0.0-20220217190341-94cf65c2a29f | |
| @@ -64,7 +64,7 @@ require ( | |
| github.com/elastic/gosigar v0.14.1 | |
| github.com/emicklei/dot v0.15.0 | |
| github.com/facebookgo/clock v0.0.0-20150410010913-600d898af40a | |
| - github.com/fraugster/parquet-go v0.10.0 | |
| + github.com/fraugster/parquet-go v0.6.1 | |
| github.com/fsnotify/fsnotify v1.5.1 | |
| github.com/getsentry/sentry-go v0.12.0 | |
| github.com/ghemawat/stream v0.0.0-20171120220530-696b145b53b9 | |
| @@ -188,7 +188,7 @@ require ( | |
| github.com/PuerkitoBio/urlesc v0.0.0-20170810143723-de5bf2ad4578 // indirect | |
| github.com/abbot/go-http-auth v0.4.1-0.20181019201920-860ed7f246ff // indirect | |
| github.com/alexbrainman/sspi v0.0.0-20180613141037-e580b900e9f5 // indirect | |
| - github.com/apache/thrift v0.16.0 // indirect | |
| + github.com/apache/thrift v0.15.0 // indirect | |
| github.com/asaskevich/govalidator v0.0.0-20200907205600-7a23bdc65eef // indirect | |
| github.com/aws/aws-sdk-go-v2/credentials v1.4.2 // indirect | |
| github.com/aws/aws-sdk-go-v2/feature/ec2/imds v1.5.1 // indirect | |
| @@ -268,7 +268,7 @@ require ( | |
| github.com/magiconair/properties v1.8.5 // indirect | |
| github.com/mailru/easyjson v0.7.6 // indirect | |
| github.com/mattn/go-ieproxy v0.0.1 // indirect | |
| - github.com/mattn/go-runewidth v0.0.10 // indirect | |
| + github.com/mattn/go-runewidth v0.0.7 // indirect | |
| github.com/matttproud/golang_protobuf_extensions v1.0.2-0.20181231171920-c182affec369 // indirect | |
| github.com/minio/md5-simd v1.1.2 // indirect | |
| github.com/minio/minio-go/v7 v7.0.21 // indirect | |
| @@ -293,7 +293,6 @@ require ( | |
| github.com/pquerna/cachecontrol v0.0.0-20200921180117-858c6e7e6b7e // indirect | |
| github.com/prometheus/procfs v0.7.3 // indirect | |
| github.com/pseudomuto/protokit v0.2.0 // indirect | |
| - github.com/rivo/uniseg v0.1.0 // indirect | |
| github.com/rogpeppe/go-internal v1.8.1 // indirect | |
| github.com/rs/xid v1.3.0 // indirect | |
| github.com/russross/blackfriday v1.6.0 // indirect | |
| diff --git a/go.sum b/go.sum | |
| index b9033e86fc..7c583ce7ce 100644 | |
| --- a/go.sum | |
| +++ b/go.sum | |
| @@ -266,10 +266,8 @@ github.com/apache/arrow/go/arrow v0.0.0-20200923215132-ac86123a3f01/go.mod h1:QN | |
| github.com/apache/thrift v0.0.0-20151001171628-53dd39833a08/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ= | |
| github.com/apache/thrift v0.12.0/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ= | |
| github.com/apache/thrift v0.13.0/go.mod h1:cp2SuWMxlEZw2r+iP2GNCdIi4C1qmUzdZFSVb+bacwQ= | |
| +github.com/apache/thrift v0.15.0 h1:aGvdaR0v1t9XLgjtBYwxcBvBOTMqClzwE26CHOgjW1Y= | |
| github.com/apache/thrift v0.15.0/go.mod h1:PHK3hniurgQaNMZYaCLEqXKsYK8upmhPbmdP2FXSqgU= | |
| -github.com/apache/thrift v0.16.0 h1:qEy6UW60iVOlUy+b9ZR0d5WzUWYGOo4HfopoyBaNmoY= | |
| -github.com/apache/thrift v0.16.0/go.mod h1:PHK3hniurgQaNMZYaCLEqXKsYK8upmhPbmdP2FXSqgU= | |
| -github.com/araddon/dateparse v0.0.0-20210429162001-6b43995a97de/go.mod h1:DCaWoUhZrYW9p1lxo/cm8EmUOOzAPSEZNGF2DK1dJgw= | |
| github.com/armon/circbuf v0.0.0-20150827004946-bbbad097214e h1:QEF07wC0T1rKkctt1RINW/+RMTVmiwxETico2l3gxJA= | |
| github.com/armon/circbuf v0.0.0-20150827004946-bbbad097214e/go.mod h1:3U/XgcO3hCbHZ8TKRvWD2dDTCfh9M9ya+I9JpbB7O8o= | |
| github.com/armon/consul-api v0.0.0-20180202201655-eb2c6b5be1b6/go.mod h1:grANhF5doyWs3UAsr3K4I6qtAmlQcZDesFNEHPZAzj8= | |
| @@ -438,8 +436,8 @@ github.com/cockroachdb/logtags v0.0.0-20211118104740-dabe8e521a4f h1:6jduT9Hfc0n | |
| github.com/cockroachdb/logtags v0.0.0-20211118104740-dabe8e521a4f/go.mod h1:Vz9DsVWQQhf3vs21MhPMZpMGSht7O/2vFW2xusFUVOs= | |
| github.com/cockroachdb/panicparse/v2 v2.0.0-20211103220158-604c82a44f1e h1:FrERdkPlRj+v7fc+PGpey3GUiDGuTR5CsmLCA54YJ8I= | |
| github.com/cockroachdb/panicparse/v2 v2.0.0-20211103220158-604c82a44f1e/go.mod h1:pMxsKyCewnV3xPaFvvT9NfwvDTcIx2Xqg0qL5Gq0SjM= | |
| -github.com/cockroachdb/pebble v0.0.0-20220307192532-e2b7bb844759 h1:PKDkU1nARLt2TL99CCEukKJIEUo2cgt9AEVlrdtzfuM= | |
| -github.com/cockroachdb/pebble v0.0.0-20220307192532-e2b7bb844759/go.mod h1:buxOO9GBtOcq1DiXDpIPYrmxY020K2A8lOrwno5FetU= | |
| +github.com/cockroachdb/pebble v0.0.0-20220227235451-40d39da505a5 h1:6ZsiW1sWGEsx2kDq98bdoDfdDeO2IgfI4e2FxUQwkdk= | |
| +github.com/cockroachdb/pebble v0.0.0-20220227235451-40d39da505a5/go.mod h1:buxOO9GBtOcq1DiXDpIPYrmxY020K2A8lOrwno5FetU= | |
| github.com/cockroachdb/redact v1.0.8/go.mod h1:BVNblN9mBWFyMyqK1k3AAiSxhvhfK2oOZZ2lK+dpvRg= | |
| github.com/cockroachdb/redact v1.1.3 h1:AKZds10rFSIj7qADf0g46UixK8NNLwWTNdCIGS5wfSQ= | |
| github.com/cockroachdb/redact v1.1.3/go.mod h1:BVNblN9mBWFyMyqK1k3AAiSxhvhfK2oOZZ2lK+dpvRg= | |
| @@ -714,8 +712,8 @@ github.com/frankban/quicktest v1.7.3/go.mod h1:V1d2J5pfxYH6EjBAgSK7YNXcXlTWxUHdE | |
| github.com/frankban/quicktest v1.10.1/go.mod h1:z7wHrVXJKCWP1Ev7B3iy2DivmuL5uGeeJDWYz/6LLhY= | |
| github.com/frankban/quicktest v1.11.3 h1:8sXhOn0uLys67V8EsXLc6eszDs8VXWxL3iRvebPhedY= | |
| github.com/frankban/quicktest v1.11.3/go.mod h1:wRf/ReqHper53s+kmmSZizM8NamnL3IM0I9ntUbOk+k= | |
| -github.com/fraugster/parquet-go v0.10.0 h1:whX91AO3dkkOnbH9MqD53DZ3rISw+Tnnj5yiqXjSv9Q= | |
| -github.com/fraugster/parquet-go v0.10.0/go.mod h1:asQOKX0K/j+F3Xyj87kw7gKrU3yXo9M2hb8STSQKIIw= | |
| +github.com/fraugster/parquet-go v0.6.1 h1:Kks9Ibly3ZZQPRPGkS1lVmrwndBp8PxamBnDFG5jvEM= | |
| +github.com/fraugster/parquet-go v0.6.1/go.mod h1:1HGhXzpHv7CULzknVNWIY0Ihn2O3qNbD1p+aQvHWhqo= | |
| github.com/fsnotify/fsnotify v1.4.7/go.mod h1:jwhsz4b93w/PPRr/qN1Yymfu8t87LnFCMoQvtojpjFo= | |
| github.com/fsnotify/fsnotify v1.4.9/go.mod h1:znqG4EE+3YCdAaPaxE2ZRY/06pZUdp0tY4IgpuI1SZQ= | |
| github.com/fsnotify/fsnotify v1.5.1 h1:mZcQUHVQUQWoPXXtuf9yuEXKudkV2sx1E06UadKWpgI= | |
| @@ -1539,9 +1537,8 @@ github.com/mattn/go-isatty v0.0.14 h1:yVuAays6BHfxijgZPzw+3Zlu5yQgKGP2/hcQbHb7S9 | |
| github.com/mattn/go-isatty v0.0.14/go.mod h1:7GGIvUiUoEMVVmxf/4nioHXj79iQHKdU27kJ6hsGG94= | |
| github.com/mattn/go-runewidth v0.0.2/go.mod h1:LwmH8dsx7+W8Uxz3IHJYH5QSwggIsqBzpuz5H//U1FU= | |
| github.com/mattn/go-runewidth v0.0.3/go.mod h1:LwmH8dsx7+W8Uxz3IHJYH5QSwggIsqBzpuz5H//U1FU= | |
| +github.com/mattn/go-runewidth v0.0.7 h1:Ei8KR0497xHyKJPAv59M1dkC+rOZCMBJ+t3fZ+twI54= | |
| github.com/mattn/go-runewidth v0.0.7/go.mod h1:H031xJmbD/WCDINGzjvQ9THkh0rPKHF+m2gUSrubnMI= | |
| -github.com/mattn/go-runewidth v0.0.10 h1:CoZ3S2P7pvtP45xOtBw+/mDL2z0RKI576gSkzRRpdGg= | |
| -github.com/mattn/go-runewidth v0.0.10/go.mod h1:RAqKPSqVFrSLVXbA8x7dzmKdmGzieGRCM46jaSJTDAk= | |
| github.com/mattn/go-shellwords v1.0.3/go.mod h1:3xCvwCdWdlDJUrvuMn7Wuy9eWs4pE8vqg+NOMyg4B2o= | |
| github.com/mattn/go-sqlite3 v1.11.0/go.mod h1:FPy6KqzDD04eiIsT53CuJW3U88zkxoIYsOqkbpncsNc= | |
| github.com/mattn/go-sqlite3 v1.14.0/go.mod h1:JIl7NbARA7phWnGvh0LKTyg7S9BA+6gx71ShQilpsus= | |
| @@ -1876,8 +1873,6 @@ github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475 h1:N/ElC8H3+5X | |
| github.com/rcrowley/go-metrics v0.0.0-20201227073835-cf1acfcdf475/go.mod h1:bCqnVzQkZxMG4s8nGwiZ5l3QUCyqpo9Y+/ZMZ9VjZe4= | |
| github.com/remyoudompheng/bigfft v0.0.0-20170806203942-52369c62f446/go.mod h1:uYEyJGbgTkfkS4+E/PavXkNJcbFIpEtjt2B0KDQ5+9M= | |
| github.com/retailnext/hllpp v1.0.1-0.20180308014038-101a6d2f8b52/go.mod h1:RDpi1RftBQPUCDRw6SmxeaREsAaRKnOclghuzp/WRzc= | |
| -github.com/rivo/uniseg v0.1.0 h1:+2KBaVoUmb9XzDsrx/Ct0W/EYOSFf/nWTauy++DprtY= | |
| -github.com/rivo/uniseg v0.1.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc= | |
| github.com/robertkrimen/godocdown v0.0.0-20130622164427-0bfa04905481/go.mod h1:C9WhFzY47SzYBIvzFqSvHIR6ROgDo4TtdTuRaOMjF/s= | |
| github.com/robfig/cron/v3 v3.0.1 h1:WdRxkvbJztn8LMz/QEvLN5sBU+xKpSqwwUO1Pjr4qDs= | |
| github.com/robfig/cron/v3 v3.0.1/go.mod h1:eQICP3HwyT7UooqI/z+Ov+PtYAWygg1TEWWzGIFLtro= | |
| @@ -1918,7 +1913,6 @@ github.com/savsgio/gotils v0.0.0-20210921075833-21a6215cb0e4/go.mod h1:oejLrk1Y/ | |
| github.com/scaleway/scaleway-sdk-go v1.0.0-beta.7.0.20210223165440-c65ae3540d44/go.mod h1:CJJ5VAbozOl0yEw7nHB9+7BXTJbIn6h7W+f6Gau5IP8= | |
| github.com/schollz/closestmatch v2.1.0+incompatible/go.mod h1:RtP1ddjLong6gTkbtmuhtR2uUrrJOpYzYRvbcPAid+g= | |
| github.com/sclevine/agouti v3.0.0+incompatible/go.mod h1:b4WX9W9L1sfQKXeJf1mUTLZKJ48R1S7H23Ji7oFO5Bw= | |
| -github.com/scylladb/termtables v0.0.0-20191203121021-c4c0b6d42ff4/go.mod h1:C1a7PQSMz9NShzorzCiG2fk9+xuCgLkPeCvMHYR2OWg= | |
| github.com/sean-/seed v0.0.0-20170313163322-e2103e2c3529/go.mod h1:DxrIzT+xaE7yg65j358z/aeFdxmN0P9QXhEzd20vsDc= | |
| github.com/seccomp/libseccomp-golang v0.9.1/go.mod h1:GbW5+tmTXfcxTToHLXlScSlAvWlF4P2Ca7zGrPiEpWo= | |
| github.com/sectioneight/md-to-godoc v0.0.0-20161108233149-55e43be6c335/go.mod h1:lPZq22klO8la1kyImIDhrGytugMV0TsrsZB55a+xxI0= | |
| diff --git a/pkg/BUILD.bazel b/pkg/BUILD.bazel | |
| index 006b4a8eaa..77752a755a 100644 | |
| Binary files a/pkg/BUILD.bazel and b/pkg/BUILD.bazel differ | |
| diff --git a/pkg/base/testing_knobs.go b/pkg/base/testing_knobs.go | |
| index 15abf6afe8..6e72ca49c3 100644 | |
| --- a/pkg/base/testing_knobs.go | |
| +++ b/pkg/base/testing_knobs.go | |
| @@ -19,34 +19,32 @@ type ModuleTestingKnobs interface { | |
| // TestingKnobs contains facilities for controlling various parts of the | |
| // system for testing. | |
| type TestingKnobs struct { | |
| - Store ModuleTestingKnobs | |
| - KVClient ModuleTestingKnobs | |
| - RangeFeed ModuleTestingKnobs | |
| - SQLExecutor ModuleTestingKnobs | |
| - SQLLeaseManager ModuleTestingKnobs | |
| - SQLSchemaChanger ModuleTestingKnobs | |
| - SQLDeclarativeSchemaChanger ModuleTestingKnobs | |
| - SQLTypeSchemaChanger ModuleTestingKnobs | |
| - GCJob ModuleTestingKnobs | |
| - PGWireTestingKnobs ModuleTestingKnobs | |
| - StartupMigrationManager ModuleTestingKnobs | |
| - DistSQL ModuleTestingKnobs | |
| - SQLEvalContext ModuleTestingKnobs | |
| - NodeLiveness ModuleTestingKnobs | |
| - Server ModuleTestingKnobs | |
| - TenantTestingKnobs ModuleTestingKnobs | |
| - JobsTestingKnobs ModuleTestingKnobs | |
| - BackupRestore ModuleTestingKnobs | |
| - TTL ModuleTestingKnobs | |
| - Streaming ModuleTestingKnobs | |
| - MigrationManager ModuleTestingKnobs | |
| - IndexUsageStatsKnobs ModuleTestingKnobs | |
| - SQLStatsKnobs ModuleTestingKnobs | |
| - SpanConfig ModuleTestingKnobs | |
| - SQLLivenessKnobs ModuleTestingKnobs | |
| - TelemetryLoggingKnobs ModuleTestingKnobs | |
| - DialerKnobs ModuleTestingKnobs | |
| - ProtectedTS ModuleTestingKnobs | |
| - CapturedIndexUsageStatsKnobs ModuleTestingKnobs | |
| - AdmissionControl ModuleTestingKnobs | |
| + Store ModuleTestingKnobs | |
| + KVClient ModuleTestingKnobs | |
| + RangeFeed ModuleTestingKnobs | |
| + SQLExecutor ModuleTestingKnobs | |
| + SQLLeaseManager ModuleTestingKnobs | |
| + SQLSchemaChanger ModuleTestingKnobs | |
| + SQLDeclarativeSchemaChanger ModuleTestingKnobs | |
| + SQLTypeSchemaChanger ModuleTestingKnobs | |
| + GCJob ModuleTestingKnobs | |
| + PGWireTestingKnobs ModuleTestingKnobs | |
| + StartupMigrationManager ModuleTestingKnobs | |
| + DistSQL ModuleTestingKnobs | |
| + SQLEvalContext ModuleTestingKnobs | |
| + NodeLiveness ModuleTestingKnobs | |
| + Server ModuleTestingKnobs | |
| + TenantTestingKnobs ModuleTestingKnobs | |
| + JobsTestingKnobs ModuleTestingKnobs | |
| + BackupRestore ModuleTestingKnobs | |
| + TTL ModuleTestingKnobs | |
| + Streaming ModuleTestingKnobs | |
| + MigrationManager ModuleTestingKnobs | |
| + IndexUsageStatsKnobs ModuleTestingKnobs | |
| + SQLStatsKnobs ModuleTestingKnobs | |
| + SpanConfig ModuleTestingKnobs | |
| + SQLLivenessKnobs ModuleTestingKnobs | |
| + TelemetryLoggingKnobs ModuleTestingKnobs | |
| + DialerKnobs ModuleTestingKnobs | |
| + ProtectedTS ModuleTestingKnobs | |
| } | |
| diff --git a/pkg/bench/rttanalysis/BUILD.bazel b/pkg/bench/rttanalysis/BUILD.bazel | |
| index 797bdf94d0..db98f19200 100644 | |
| --- a/pkg/bench/rttanalysis/BUILD.bazel | |
| +++ b/pkg/bench/rttanalysis/BUILD.bazel | |
| @@ -49,7 +49,6 @@ go_test( | |
| ], | |
| data = glob(["testdata/**"]), | |
| embed = [":rttanalysis"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/security", | |
| diff --git a/pkg/bench/rttanalysis/testdata/benchmark_expectations b/pkg/bench/rttanalysis/testdata/benchmark_expectations | |
| index 10b154c3a3..311a9e95d1 100644 | |
| --- a/pkg/bench/rttanalysis/testdata/benchmark_expectations | |
| +++ b/pkg/bench/rttanalysis/testdata/benchmark_expectations | |
| @@ -50,8 +50,8 @@ exp,benchmark | |
| 19,Grant/grant_all_on_1_table | |
| 19,Grant/grant_all_on_2_tables | |
| 19,Grant/grant_all_on_3_tables | |
| -13,GrantRole/grant_1_role | |
| -15,GrantRole/grant_2_roles | |
| +17,GrantRole/grant_1_role | |
| +20,GrantRole/grant_2_roles | |
| 2,ORMQueries/activerecord_type_introspection_query | |
| 2,ORMQueries/django_table_introspection_1_table | |
| 2,ORMQueries/django_table_introspection_4_tables | |
| @@ -79,8 +79,8 @@ exp,benchmark | |
| 19,Revoke/revoke_all_on_1_table | |
| 19,Revoke/revoke_all_on_2_tables | |
| 19,Revoke/revoke_all_on_3_tables | |
| -13,RevokeRole/revoke_1_role | |
| -15,RevokeRole/revoke_2_roles | |
| +16,RevokeRole/revoke_1_role | |
| +18,RevokeRole/revoke_2_roles | |
| 1,SystemDatabaseQueries/select_system.users_with_empty_database_Name | |
| 1,SystemDatabaseQueries/select_system.users_with_schema_Name | |
| 2,SystemDatabaseQueries/select_system.users_without_schema_Name | |
| diff --git a/pkg/build/BUILD.bazel b/pkg/build/BUILD.bazel | |
| index 2fdb50f105..428ae5f446 100644 | |
| --- a/pkg/build/BUILD.bazel | |
| +++ b/pkg/build/BUILD.bazel | |
| @@ -15,11 +15,10 @@ go_library( | |
| visibility = ["//visibility:public"], | |
| x_defs = { | |
| "github.com/cockroachdb/cockroach/pkg/build.cgoTargetTriple": "{STABLE_BUILD_TARGET_TRIPLE}", | |
| - "github.com/cockroachdb/cockroach/pkg/build.channel": "{STABLE_BUILD_CHANNEL}", | |
| - "github.com/cockroachdb/cockroach/pkg/build.rev": "{BUILD_REV}", | |
| - "github.com/cockroachdb/cockroach/pkg/build.tag": "{BUILD_TAG}", | |
| - "github.com/cockroachdb/cockroach/pkg/build.typ": "{STABLE_BUILD_TYPE}", | |
| - "github.com/cockroachdb/cockroach/pkg/build.utcTime": "{BUILD_UTCTIME}", | |
| + "github.com/cockroachdb/cockroach/pkg/build.typ": "{STABLE_BUILD_GIT_BUILD_TYPE}", | |
| + "github.com/cockroachdb/cockroach/pkg/build.rev": "{BUILD_GIT_COMMIT}", | |
| + "github.com/cockroachdb/cockroach/pkg/build.tag": "{BUILD_GIT_TAG}", | |
| + "github.com/cockroachdb/cockroach/pkg/build.utcTime": "{BUILD_GIT_UTCTIME}", | |
| }, | |
| deps = [ | |
| "//pkg/util/envutil", | |
| diff --git a/pkg/ccl/backupccl/BUILD.bazel b/pkg/ccl/backupccl/BUILD.bazel | |
| index e735fd3302..5d9da862bb 100644 | |
| --- a/pkg/ccl/backupccl/BUILD.bazel | |
| +++ b/pkg/ccl/backupccl/BUILD.bazel | |
| @@ -9,7 +9,6 @@ go_library( | |
| "backup.go", | |
| "backup_destination.go", | |
| "backup_job.go", | |
| - "backup_metadata.go", | |
| "backup_planning.go", | |
| "backup_planning_tenant.go", | |
| "backup_processor.go", | |
| @@ -121,7 +120,6 @@ go_library( | |
| "//pkg/util/hlc", | |
| "//pkg/util/interval", | |
| "//pkg/util/ioctx", | |
| - "//pkg/util/json", | |
| "//pkg/util/log", | |
| "//pkg/util/log/eventpb", | |
| "//pkg/util/metric", | |
| @@ -154,7 +152,6 @@ go_test( | |
| "backup_cloud_test.go", | |
| "backup_destination_test.go", | |
| "backup_intents_test.go", | |
| - "backup_metadata_test.go", | |
| "backup_rand_test.go", | |
| "backup_tenant_test.go", | |
| "backup_test.go", | |
| @@ -251,7 +248,6 @@ go_test( | |
| "//pkg/util/ctxgroup", | |
| "//pkg/util/encoding", | |
| "//pkg/util/hlc", | |
| - "//pkg/util/ioctx", | |
| "//pkg/util/leaktest", | |
| "//pkg/util/log", | |
| "//pkg/util/mon", | |
| diff --git a/pkg/ccl/backupccl/backup_job.go b/pkg/ccl/backupccl/backup_job.go | |
| index 46a4674206..ef8166e451 100644 | |
| --- a/pkg/ccl/backupccl/backup_job.go | |
| +++ b/pkg/ccl/backupccl/backup_job.go | |
| @@ -165,7 +165,6 @@ func backup( | |
| planCtx, | |
| execCtx, | |
| dsp, | |
| - int64(job.ID()), | |
| spans, | |
| introducedSpans, | |
| pkIDs, | |
| @@ -332,16 +331,6 @@ func backup( | |
| return roachpb.RowCount{}, err | |
| } | |
| - if writeMetadataSST.Get(&settings.SV) { | |
| - if err := writeBackupMetadataSST(ctx, defaultStore, encryption, backupManifest, tableStatistics); err != nil { | |
| - err = errors.Wrap(err, "writing forward-compat metadata sst") | |
| - if !build.IsRelease() { | |
| - return roachpb.RowCount{}, err | |
| - } | |
| - log.Warningf(ctx, "%+v", err) | |
| - } | |
| - } | |
| - | |
| return backupManifest.EntryCounts, nil | |
| } | |
| @@ -481,6 +470,20 @@ func (b *backupResumer) Resume(ctx context.Context, execCtx interface{}) error { | |
| } | |
| } | |
| + ptsID := details.ProtectedTimestampRecord | |
| + if ptsID != nil && !b.testingKnobs.ignoreProtectedTimestamps { | |
| + resumerSpan.RecordStructured(&types.StringValue{Value: "verifying protected timestamp"}) | |
| + if err := p.ExecCfg().ProtectedTimestampProvider.Verify(ctx, *ptsID); err != nil { | |
| + if errors.Is(err, protectedts.ErrNotExists) { | |
| + // No reason to return an error which might cause problems if it doesn't | |
| + // seem to exist. | |
| + log.Warningf(ctx, "failed to release protected which seems not to exist: %v", err) | |
| + } else { | |
| + return err | |
| + } | |
| + } | |
| + } | |
| + | |
| storageByLocalityKV := make(map[string]*roachpb.ExternalStorage) | |
| for kv, uri := range details.URIsByLocalityKV { | |
| conf, err := cloud.ExternalStorageConfFromURI(uri, p.User()) | |
| @@ -578,7 +581,7 @@ func (b *backupResumer) Resume(ctx context.Context, execCtx interface{}) error { | |
| return err | |
| } | |
| - if details.ProtectedTimestampRecord != nil && !b.testingKnobs.ignoreProtectedTimestamps { | |
| + if ptsID != nil && !b.testingKnobs.ignoreProtectedTimestamps { | |
| if err := p.ExecCfg().DB.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| details := b.job.Details().(jobspb.BackupDetails) | |
| return releaseProtectedTimestamp(ctx, txn, p.ExecCfg().ProtectedTimestampProvider, | |
| diff --git a/pkg/ccl/backupccl/backup_metadata.go b/pkg/ccl/backupccl/backup_metadata.go | |
| deleted file mode 100644 | |
| index fee3e840d6..0000000000 | |
| --- a/pkg/ccl/backupccl/backup_metadata.go | |
| +++ /dev/null | |
| @@ -1,1265 +0,0 @@ | |
| -// Copyright 2021 The Cockroach Authors. | |
| -// | |
| -// Licensed as a CockroachDB Enterprise file under the Cockroach Community | |
| -// License (the "License"); you may not use this file except in compliance with | |
| -// the License. You may obtain a copy of the License at | |
| -// | |
| -// https://github.com/cockroachdb/cockroach/blob/master/licenses/CCL.txt | |
| - | |
| -package backupccl | |
| - | |
| -import ( | |
| - "bytes" | |
| - "context" | |
| - "fmt" | |
| - "io" | |
| - "sort" | |
| - "strings" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/ccl/storageccl" | |
| - "github.com/cockroachdb/cockroach/pkg/cloud" | |
| - "github.com/cockroachdb/cockroach/pkg/jobs/jobspb" | |
| - "github.com/cockroachdb/cockroach/pkg/keys" | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/catalog/descpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/protoreflect" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/stats" | |
| - "github.com/cockroachdb/cockroach/pkg/storage" | |
| - "github.com/cockroachdb/cockroach/pkg/util/encoding" | |
| - "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| - "github.com/cockroachdb/cockroach/pkg/util/json" | |
| - "github.com/cockroachdb/cockroach/pkg/util/protoutil" | |
| - "github.com/cockroachdb/errors" | |
| -) | |
| - | |
| -const ( | |
| - metadataSSTName = "metadata.sst" | |
| - fileInfoPath = "fileinfo.sst" | |
| - sstBackupKey = "backup" | |
| - sstDescsPrefix = "desc/" | |
| - sstFilesPrefix = "file/" | |
| - sstNamesPrefix = "name/" | |
| - sstSpansPrefix = "span/" | |
| - sstStatsPrefix = "stats/" | |
| - sstTenantsPrefix = "tenant/" | |
| -) | |
| - | |
| -func writeBackupMetadataSST( | |
| - ctx context.Context, | |
| - dest cloud.ExternalStorage, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - manifest *BackupManifest, | |
| - stats []*stats.TableStatisticProto, | |
| -) error { | |
| - var w io.WriteCloser | |
| - ctx, cancel := context.WithCancel(ctx) | |
| - defer func() { | |
| - cancel() // cancel before Close() to abort write on err returns. | |
| - if w != nil { | |
| - w.Close() | |
| - } | |
| - }() | |
| - | |
| - w, err := makeWriter(ctx, dest, metadataSSTName, enc) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := constructMetadataSST(ctx, dest, enc, w, manifest, stats); err != nil { | |
| - return err | |
| - } | |
| - | |
| - // Explicitly close to flush and check for errors do so before defer's cancel | |
| - // which would abort. Then nil out w to avoid defer double-closing. | |
| - err = w.Close() | |
| - w = nil | |
| - return err | |
| -} | |
| - | |
| -func makeWriter( | |
| - ctx context.Context, | |
| - dest cloud.ExternalStorage, | |
| - filename string, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| -) (io.WriteCloser, error) { | |
| - w, err := dest.Writer(ctx, filename) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - if enc != nil { | |
| - key, err := getEncryptionKey(ctx, enc, dest.Settings(), dest.ExternalIOConf()) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - encW, err := storageccl.EncryptingWriter(w, key) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - w = encW | |
| - } | |
| - return w, nil | |
| -} | |
| - | |
| -func constructMetadataSST( | |
| - ctx context.Context, | |
| - dest cloud.ExternalStorage, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - w io.Writer, | |
| - m *BackupManifest, | |
| - stats []*stats.TableStatisticProto, | |
| -) error { | |
| - // TODO(dt): use a seek-optimized SST writer instead. | |
| - sst := storage.MakeBackupSSTWriter(ctx, dest.Settings(), w) | |
| - defer sst.Close() | |
| - | |
| - // The following steps must be done in-order, by key prefix. | |
| - | |
| - if err := writeManifestToMetadata(ctx, sst, m); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeDescsToMetadata(ctx, sst, m); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeFilesToMetadata(ctx, sst, m, dest, enc, fileInfoPath); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeNamesToMetadata(ctx, sst, m); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeSpansToMetadata(ctx, sst, m); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeStatsToMetadata(ctx, sst, stats); err != nil { | |
| - return err | |
| - } | |
| - | |
| - if err := writeTenantsToMetadata(ctx, sst, m); err != nil { | |
| - return err | |
| - } | |
| - | |
| - return sst.Finish() | |
| -} | |
| - | |
| -func writeManifestToMetadata(ctx context.Context, sst storage.SSTWriter, m *BackupManifest) error { | |
| - info := *m | |
| - info.Descriptors = nil | |
| - info.DescriptorChanges = nil | |
| - info.Files = nil | |
| - info.Spans = nil | |
| - info.StatisticsFilenames = nil | |
| - info.IntroducedSpans = nil | |
| - info.Tenants = nil | |
| - | |
| - b, err := protoutil.Marshal(&info) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - return sst.PutUnversioned(roachpb.Key(sstBackupKey), b) | |
| -} | |
| - | |
| -func writeDescsToMetadata(ctx context.Context, sst storage.SSTWriter, m *BackupManifest) error { | |
| - // Add descriptors from revisions if available, Descriptors if not. | |
| - if len(m.DescriptorChanges) > 0 { | |
| - sort.Slice(m.DescriptorChanges, func(i, j int) bool { | |
| - if m.DescriptorChanges[i].ID < m.DescriptorChanges[j].ID { | |
| - return true | |
| - } else if m.DescriptorChanges[i].ID == m.DescriptorChanges[j].ID { | |
| - return !m.DescriptorChanges[i].Time.Less(m.DescriptorChanges[j].Time) | |
| - } | |
| - return false | |
| - }) | |
| - for _, i := range m.DescriptorChanges { | |
| - k := encodeDescSSTKey(i.ID) | |
| - var b []byte | |
| - if i.Desc != nil { | |
| - t, _, _, _ := descpb.FromDescriptor(i.Desc) | |
| - if t == nil || !t.Dropped() { | |
| - bytes, err := protoutil.Marshal(i.Desc) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - b = bytes | |
| - } | |
| - } | |
| - if err := sst.PutMVCC(storage.MVCCKey{Key: k, Timestamp: i.Time}, b); err != nil { | |
| - return err | |
| - } | |
| - | |
| - } | |
| - } else { | |
| - sort.Slice(m.Descriptors, func(i, j int) bool { | |
| - return descID(m.Descriptors[i]) < descID(m.Descriptors[j]) | |
| - }) | |
| - for _, i := range m.Descriptors { | |
| - id := descID(i) | |
| - k := encodeDescSSTKey(id) | |
| - b, err := protoutil.Marshal(&i) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - | |
| - // Put descriptors at start time. For non-rev backups this timestamp | |
| - // doesn't matter. For the special case where there were no descriptor | |
| - // changes in an incremental backup, it's helpful to have existing | |
| - // descriptors at the start time, so we don't have to look back further | |
| - // than the very last backup. | |
| - if m.StartTime.IsEmpty() { | |
| - if err := sst.PutUnversioned(k, b); err != nil { | |
| - return err | |
| - } | |
| - } else { | |
| - if err := sst.PutMVCC(storage.MVCCKey{Key: k, Timestamp: m.StartTime}, b); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -func writeFilesToMetadata( | |
| - ctx context.Context, | |
| - sst storage.SSTWriter, | |
| - m *BackupManifest, | |
| - dest cloud.ExternalStorage, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - fileInfoPath string, | |
| -) error { | |
| - w, err := makeWriter(ctx, dest, fileInfoPath, enc) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - defer w.Close() | |
| - fileSST := storage.MakeBackupSSTWriter(ctx, dest.Settings(), w) | |
| - defer fileSST.Close() | |
| - | |
| - // Sort and write all of the files into a single file info SST. | |
| - sort.Slice(m.Files, func(i, j int) bool { | |
| - cmp := m.Files[i].Span.Key.Compare(m.Files[j].Span.Key) | |
| - return cmp < 0 || (cmp == 0 && strings.Compare(m.Files[i].Path, m.Files[j].Path) < 0) | |
| - }) | |
| - | |
| - for _, i := range m.Files { | |
| - b, err := protoutil.Marshal(&i) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := fileSST.PutUnversioned(encodeFileSSTKey(i.Span.Key, i.Path), b); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - | |
| - err = fileSST.Finish() | |
| - if err != nil { | |
| - return err | |
| - } | |
| - err = w.Close() | |
| - if err != nil { | |
| - return err | |
| - } | |
| - | |
| - // Write the file info into the main metadata SST. | |
| - return sst.PutUnversioned(encodeFilenameSSTKey(fileInfoPath), nil) | |
| -} | |
| - | |
| -type name struct { | |
| - parent, parentSchema descpb.ID | |
| - name string | |
| - id descpb.ID | |
| - ts hlc.Timestamp | |
| -} | |
| - | |
| -type namespace []name | |
| - | |
| -func (a namespace) Len() int { return len(a) } | |
| -func (a namespace) Swap(i, j int) { a[i], a[j] = a[j], a[i] } | |
| -func (a namespace) Less(i, j int) bool { | |
| - if a[i].parent == a[j].parent { | |
| - if a[i].parentSchema == a[j].parentSchema { | |
| - cmp := strings.Compare(a[i].name, a[j].name) | |
| - return cmp < 0 || (cmp == 0 && (a[i].ts.IsEmpty() || a[j].ts.Less(a[i].ts))) | |
| - } | |
| - return a[i].parentSchema < a[j].parentSchema | |
| - } | |
| - return a[i].parent < a[j].parent | |
| -} | |
| - | |
| -func writeNamesToMetadata(ctx context.Context, sst storage.SSTWriter, m *BackupManifest) error { | |
| - revs := m.DescriptorChanges | |
| - if len(revs) == 0 { | |
| - revs = make([]BackupManifest_DescriptorRevision, len(m.Descriptors)) | |
| - for i := range m.Descriptors { | |
| - revs[i].Desc = &m.Descriptors[i] | |
| - revs[i].Time = m.EndTime | |
| - revs[i].ID = descID(m.Descriptors[i]) | |
| - } | |
| - } | |
| - | |
| - names := make(namespace, len(revs)) | |
| - | |
| - for i, rev := range revs { | |
| - names[i].id = rev.ID | |
| - names[i].ts = rev.Time | |
| - tb, db, typ, sc := descpb.FromDescriptor(rev.Desc) | |
| - if db != nil { | |
| - names[i].name = db.Name | |
| - } else if sc != nil { | |
| - names[i].name = sc.Name | |
| - names[i].parent = sc.ParentID | |
| - } else if tb != nil { | |
| - names[i].name = tb.Name | |
| - names[i].parent = tb.ParentID | |
| - names[i].parentSchema = keys.PublicSchemaID | |
| - if s := tb.UnexposedParentSchemaID; s != descpb.InvalidID { | |
| - names[i].parentSchema = s | |
| - } | |
| - if tb.Dropped() { | |
| - names[i].id = 0 | |
| - } | |
| - } else if typ != nil { | |
| - names[i].name = typ.Name | |
| - names[i].parent = typ.ParentID | |
| - names[i].parentSchema = typ.ParentSchemaID | |
| - } | |
| - } | |
| - sort.Sort(names) | |
| - | |
| - for i, rev := range names { | |
| - if i > 0 { | |
| - prev := names[i-1] | |
| - prev.ts = rev.ts | |
| - if prev == rev { | |
| - continue | |
| - } | |
| - } | |
| - k := encodeNameSSTKey(rev.parent, rev.parentSchema, rev.name) | |
| - v := encoding.EncodeUvarintAscending(nil, uint64(rev.id)) | |
| - if err := sst.PutMVCC(storage.MVCCKey{Key: k, Timestamp: rev.ts}, v); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - | |
| - return nil | |
| -} | |
| - | |
| -func writeSpansToMetadata(ctx context.Context, sst storage.SSTWriter, m *BackupManifest) error { | |
| - sort.Sort(roachpb.Spans(m.Spans)) | |
| - sort.Sort(roachpb.Spans(m.IntroducedSpans)) | |
| - | |
| - for i, j := 0, 0; i < len(m.Spans) || j < len(m.IntroducedSpans); { | |
| - var sp roachpb.Span | |
| - var ts hlc.Timestamp | |
| - | |
| - // Merge spans and introduced spans into one series of spans where the ts on | |
| - // each is 0 if it was introduced or the backup start time otherwise. | |
| - if j >= len(m.IntroducedSpans) { | |
| - sp = m.Spans[i] | |
| - ts = m.StartTime | |
| - i++ | |
| - } else if i >= len(m.Spans) { | |
| - sp = m.IntroducedSpans[j] | |
| - ts = hlc.Timestamp{} | |
| - j++ | |
| - } else { | |
| - cmp := m.Spans[i].Key.Compare(m.IntroducedSpans[j].Key) | |
| - if cmp < 0 { | |
| - sp = m.Spans[i] | |
| - ts = m.StartTime | |
| - i++ | |
| - } else { | |
| - sp = m.IntroducedSpans[j] | |
| - ts = hlc.Timestamp{} | |
| - j++ | |
| - } | |
| - } | |
| - if ts.IsEmpty() { | |
| - if err := sst.PutUnversioned(encodeSpanSSTKey(sp), nil); err != nil { | |
| - return err | |
| - } | |
| - } else { | |
| - k := storage.MVCCKey{Key: encodeSpanSSTKey(sp), Timestamp: ts} | |
| - if err := sst.PutMVCC(k, nil); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -func writeStatsToMetadata( | |
| - ctx context.Context, sst storage.SSTWriter, stats []*stats.TableStatisticProto, | |
| -) error { | |
| - sort.Slice(stats, func(i, j int) bool { | |
| - return stats[i].TableID < stats[j].TableID || (stats[i].TableID == stats[j].TableID && stats[i].StatisticID < stats[j].StatisticID) | |
| - }) | |
| - | |
| - for _, i := range stats { | |
| - b, err := protoutil.Marshal(i) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := sst.PutUnversioned(encodeStatSSTKey(i.TableID, i.StatisticID), b); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -func writeTenantsToMetadata(ctx context.Context, sst storage.SSTWriter, m *BackupManifest) error { | |
| - sort.Slice(m.Tenants, func(i, j int) bool { return m.Tenants[i].ID < m.Tenants[j].ID }) | |
| - for _, i := range m.Tenants { | |
| - b, err := protoutil.Marshal(&i) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := sst.PutUnversioned(encodeTenantSSTKey(i.ID), b); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -func descID(in descpb.Descriptor) descpb.ID { | |
| - switch i := in.Union.(type) { | |
| - case *descpb.Descriptor_Table: | |
| - return i.Table.ID | |
| - case *descpb.Descriptor_Database: | |
| - return i.Database.ID | |
| - case *descpb.Descriptor_Type: | |
| - return i.Type.ID | |
| - case *descpb.Descriptor_Schema: | |
| - return i.Schema.ID | |
| - default: | |
| - panic(fmt.Sprintf("unknown desc %T", in)) | |
| - } | |
| -} | |
| - | |
| -func deprefix(key roachpb.Key, prefix string) (roachpb.Key, error) { | |
| - if !bytes.HasPrefix(key, []byte(prefix)) { | |
| - return nil, errors.Errorf("malformed key missing expected prefix %s: %q", prefix, key) | |
| - } | |
| - return key[len(prefix):], nil | |
| -} | |
| - | |
| -func encodeDescSSTKey(id descpb.ID) roachpb.Key { | |
| - return roachpb.Key(encoding.EncodeUvarintAscending([]byte(sstDescsPrefix), uint64(id))) | |
| -} | |
| - | |
| -func decodeDescSSTKey(key roachpb.Key) (descpb.ID, error) { | |
| - key, err := deprefix(key, sstDescsPrefix) | |
| - if err != nil { | |
| - return 0, err | |
| - } | |
| - _, id, err := encoding.DecodeUvarintAscending(key) | |
| - return descpb.ID(id), err | |
| -} | |
| - | |
| -func encodeFileSSTKey(spanStart roachpb.Key, filename string) roachpb.Key { | |
| - buf := make([]byte, 0) | |
| - buf = encoding.EncodeBytesAscending(buf, spanStart) | |
| - return roachpb.Key(encoding.EncodeStringAscending(buf, filename)) | |
| -} | |
| - | |
| -func encodeFilenameSSTKey(filename string) roachpb.Key { | |
| - return encoding.EncodeStringAscending([]byte(sstFilesPrefix), filename) | |
| -} | |
| - | |
| -func decodeUnsafeFileSSTKey(key roachpb.Key) (roachpb.Key, string, error) { | |
| - key, spanStart, err := encoding.DecodeBytesAscending(key, nil) | |
| - if err != nil { | |
| - return nil, "", err | |
| - } | |
| - _, filename, err := encoding.DecodeUnsafeStringAscending(key, nil) | |
| - if err != nil { | |
| - return nil, "", err | |
| - } | |
| - return roachpb.Key(spanStart), filename, err | |
| -} | |
| - | |
| -func decodeUnsafeFileInfoSSTKey(key roachpb.Key) (string, error) { | |
| - key, err := deprefix(key, sstFilesPrefix) | |
| - if err != nil { | |
| - return "", err | |
| - } | |
| - | |
| - _, path, err := encoding.DecodeUnsafeStringAscending(key, nil) | |
| - if err != nil { | |
| - return "", err | |
| - } | |
| - return path, err | |
| -} | |
| - | |
| -func encodeNameSSTKey(parentDB, parentSchema descpb.ID, name string) roachpb.Key { | |
| - buf := []byte(sstNamesPrefix) | |
| - buf = encoding.EncodeUvarintAscending(buf, uint64(parentDB)) | |
| - buf = encoding.EncodeUvarintAscending(buf, uint64(parentSchema)) | |
| - return roachpb.Key(encoding.EncodeStringAscending(buf, name)) | |
| -} | |
| - | |
| -func decodeUnsafeNameSSTKey(key roachpb.Key) (descpb.ID, descpb.ID, string, error) { | |
| - key, err := deprefix(key, sstNamesPrefix) | |
| - if err != nil { | |
| - return 0, 0, "", err | |
| - } | |
| - key, parentID, err := encoding.DecodeUvarintAscending(key) | |
| - if err != nil { | |
| - return 0, 0, "", err | |
| - } | |
| - key, schemaID, err := encoding.DecodeUvarintAscending(key) | |
| - if err != nil { | |
| - return 0, 0, "", err | |
| - } | |
| - _, name, err := encoding.DecodeUnsafeStringAscending(key, nil) | |
| - if err != nil { | |
| - return 0, 0, "", err | |
| - } | |
| - return descpb.ID(parentID), descpb.ID(schemaID), name, nil | |
| -} | |
| - | |
| -func encodeSpanSSTKey(span roachpb.Span) roachpb.Key { | |
| - buf := encoding.EncodeBytesAscending([]byte(sstSpansPrefix), span.Key) | |
| - return roachpb.Key(encoding.EncodeBytesAscending(buf, span.EndKey)) | |
| -} | |
| - | |
| -func decodeSpanSSTKey(key roachpb.Key) (roachpb.Span, error) { | |
| - key, err := deprefix(key, sstSpansPrefix) | |
| - if err != nil { | |
| - return roachpb.Span{}, err | |
| - } | |
| - key, start, err := encoding.DecodeBytesAscending(key, nil) | |
| - if err != nil { | |
| - return roachpb.Span{}, err | |
| - } | |
| - _, end, err := encoding.DecodeBytesAscending(key, nil) | |
| - return roachpb.Span{Key: start, EndKey: end}, err | |
| -} | |
| - | |
| -func encodeStatSSTKey(id descpb.ID, statID uint64) roachpb.Key { | |
| - buf := encoding.EncodeUvarintAscending([]byte(sstStatsPrefix), uint64(id)) | |
| - return roachpb.Key(encoding.EncodeUvarintAscending(buf, statID)) | |
| -} | |
| - | |
| -func decodeStatSSTKey(key roachpb.Key) (descpb.ID, uint64, error) { | |
| - key, err := deprefix(key, sstStatsPrefix) | |
| - if err != nil { | |
| - return 0, 0, err | |
| - } | |
| - key, id, err := encoding.DecodeUvarintAscending(key) | |
| - if err != nil { | |
| - return 0, 0, err | |
| - } | |
| - _, stat, err := encoding.DecodeUvarintAscending(key) | |
| - return descpb.ID(id), stat, err | |
| -} | |
| - | |
| -func encodeTenantSSTKey(id uint64) roachpb.Key { | |
| - return encoding.EncodeUvarintAscending([]byte(sstTenantsPrefix), id) | |
| -} | |
| - | |
| -func decodeTenantSSTKey(key roachpb.Key) (uint64, error) { | |
| - key, err := deprefix(key, sstTenantsPrefix) | |
| - if err != nil { | |
| - return 0, err | |
| - } | |
| - _, id, err := encoding.DecodeUvarintAscending(key) | |
| - if err != nil { | |
| - return 0, err | |
| - } | |
| - return id, nil | |
| -} | |
| - | |
| -func pbBytesToJSON(in []byte, msg protoutil.Message) (json.JSON, error) { | |
| - if err := protoutil.Unmarshal(in, msg); err != nil { | |
| - return nil, err | |
| - } | |
| - j, err := protoreflect.MessageToJSON(msg, protoreflect.FmtFlags{}) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - return j, nil | |
| -} | |
| - | |
| -func debugDumpFileSST( | |
| - ctx context.Context, | |
| - store cloud.ExternalStorage, | |
| - fileInfoPath string, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - out func(rawKey, readableKey string, value json.JSON) error, | |
| -) error { | |
| - var encOpts *roachpb.FileEncryptionOptions | |
| - if enc != nil { | |
| - key, err := getEncryptionKey(ctx, enc, store.Settings(), store.ExternalIOConf()) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - encOpts = &roachpb.FileEncryptionOptions{Key: key} | |
| - } | |
| - iter, err := storageccl.ExternalSSTReader(ctx, store, fileInfoPath, encOpts) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - defer iter.Close() | |
| - for iter.SeekGE(storage.MVCCKey{}); ; iter.Next() { | |
| - ok, err := iter.Valid() | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if !ok { | |
| - break | |
| - } | |
| - k := iter.UnsafeKey() | |
| - spanStart, path, err := decodeUnsafeFileSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - f, err := pbBytesToJSON(iter.UnsafeValue(), &BackupManifest_File{}) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("file %s (%s)", path, spanStart.String()), f); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - | |
| - return nil | |
| -} | |
| - | |
| -// DebugDumpMetadataSST is for debugging a metadata SST. | |
| -func DebugDumpMetadataSST( | |
| - ctx context.Context, | |
| - store cloud.ExternalStorage, | |
| - path string, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - out func(rawKey, readableKey string, value json.JSON) error, | |
| -) error { | |
| - var encOpts *roachpb.FileEncryptionOptions | |
| - if enc != nil { | |
| - key, err := getEncryptionKey(ctx, enc, store.Settings(), store.ExternalIOConf()) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - encOpts = &roachpb.FileEncryptionOptions{Key: key} | |
| - } | |
| - | |
| - iter, err := storageccl.ExternalSSTReader(ctx, store, path, encOpts) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - defer iter.Close() | |
| - | |
| - for iter.SeekGE(storage.MVCCKey{}); ; iter.Next() { | |
| - ok, err := iter.Valid() | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if !ok { | |
| - break | |
| - } | |
| - k := iter.UnsafeKey() | |
| - switch { | |
| - case bytes.Equal(k.Key, []byte(sstBackupKey)): | |
| - info, err := pbBytesToJSON(iter.UnsafeValue(), &BackupManifest{}) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), "backup info", info); err != nil { | |
| - return err | |
| - } | |
| - | |
| - case bytes.HasPrefix(k.Key, []byte(sstDescsPrefix)): | |
| - id, err := decodeDescSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - var desc json.JSON | |
| - if v := iter.UnsafeValue(); len(v) > 0 { | |
| - desc, err = pbBytesToJSON(v, &descpb.Descriptor{}) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("desc %d @ %v", id, k.Timestamp), desc); err != nil { | |
| - return err | |
| - } | |
| - | |
| - case bytes.HasPrefix(k.Key, []byte(sstFilesPrefix)): | |
| - p, err := decodeUnsafeFileInfoSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("file info @ %s", p), nil); err != nil { | |
| - return err | |
| - } | |
| - if err := debugDumpFileSST(ctx, store, p, enc, out); err != nil { | |
| - return err | |
| - } | |
| - case bytes.HasPrefix(k.Key, []byte(sstNamesPrefix)): | |
| - db, sc, name, err := decodeUnsafeNameSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - var id uint64 | |
| - if v := iter.UnsafeValue(); len(v) > 0 { | |
| - _, id, err = encoding.DecodeUvarintAscending(v) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - } | |
| - mapping := fmt.Sprintf("name db %d / schema %d / %q @ %v -> %d", db, sc, name, k.Timestamp, id) | |
| - if err := out(k.String(), mapping, nil); err != nil { | |
| - return err | |
| - } | |
| - | |
| - case bytes.HasPrefix(k.Key, []byte(sstSpansPrefix)): | |
| - span, err := decodeSpanSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("span %s @ %v", span, k.Timestamp), nil); err != nil { | |
| - return err | |
| - } | |
| - | |
| - case bytes.HasPrefix(k.Key, []byte(sstStatsPrefix)): | |
| - tblID, statID, err := decodeStatSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - s, err := pbBytesToJSON(iter.UnsafeValue(), &stats.TableStatisticProto{}) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("stats tbl %d, id %d", tblID, statID), s); err != nil { | |
| - return err | |
| - } | |
| - | |
| - case bytes.HasPrefix(k.Key, []byte(sstTenantsPrefix)): | |
| - id, err := decodeTenantSSTKey(k.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - i, err := pbBytesToJSON(iter.UnsafeValue(), &descpb.TenantInfo{}) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if err := out(k.String(), fmt.Sprintf("tenant %d", id), i); err != nil { | |
| - return err | |
| - } | |
| - | |
| - default: | |
| - if err := out(k.String(), "unknown", json.FromString(fmt.Sprintf("%q", iter.UnsafeValue()))); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - } | |
| - | |
| - return nil | |
| -} | |
| - | |
| -// BackupMetadata holds all of the data in BackupManifest except a few repeated | |
| -// fields such as descriptors or spans. BackupMetadata provides iterator methods | |
| -// so that the excluded fields can be accessed in a streaming manner. | |
| -type BackupMetadata struct { | |
| - BackupManifest | |
| - store cloud.ExternalStorage | |
| - enc *jobspb.BackupEncryptionOptions | |
| - filename string | |
| -} | |
| - | |
| -func newBackupMetadata( | |
| - ctx context.Context, | |
| - exportStore cloud.ExternalStorage, | |
| - sstFileName string, | |
| - encryption *jobspb.BackupEncryptionOptions, | |
| -) (*BackupMetadata, error) { | |
| - var encOpts *roachpb.FileEncryptionOptions | |
| - if encryption != nil { | |
| - key, err := getEncryptionKey(ctx, encryption, exportStore.Settings(), exportStore.ExternalIOConf()) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - encOpts = &roachpb.FileEncryptionOptions{Key: key} | |
| - } | |
| - | |
| - iter, err := storageccl.ExternalSSTReader(ctx, exportStore, sstFileName, encOpts) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - defer iter.Close() | |
| - | |
| - var sstManifest BackupManifest | |
| - iter.SeekGE(storage.MakeMVCCMetadataKey([]byte(sstBackupKey))) | |
| - ok, err := iter.Valid() | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - if !ok || !iter.UnsafeKey().Key.Equal([]byte(sstBackupKey)) { | |
| - return nil, errors.Errorf("metadata SST does not contain backup manifest") | |
| - } | |
| - | |
| - if err := protoutil.Unmarshal(iter.UnsafeValue(), &sstManifest); err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - return &BackupMetadata{BackupManifest: sstManifest, store: exportStore, enc: encryption, filename: sstFileName}, nil | |
| -} | |
| - | |
| -// SpanIterator is a simple iterator to iterate over roachpb.Spans. | |
| -type SpanIterator struct { | |
| - backing bytesIter | |
| - filter func(key storage.MVCCKey) bool | |
| - err error | |
| -} | |
| - | |
| -// SpanIter creates a new SpanIterator for the backup metadata. | |
| -func (b *BackupMetadata) SpanIter(ctx context.Context) SpanIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstSpansPrefix), b.enc, true) | |
| - return SpanIterator{ | |
| - backing: backing, | |
| - } | |
| -} | |
| - | |
| -// IntroducedSpanIter creates a new IntroducedSpanIterator for the backup metadata. | |
| -func (b *BackupMetadata) IntroducedSpanIter(ctx context.Context) SpanIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstSpansPrefix), b.enc, false) | |
| - | |
| - return SpanIterator{ | |
| - backing: backing, | |
| - filter: func(key storage.MVCCKey) bool { | |
| - return key.Timestamp == hlc.Timestamp{} | |
| - }, | |
| - } | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (si *SpanIterator) Close() { | |
| - si.backing.close() | |
| -} | |
| - | |
| -// Err returns the iterator's error | |
| -func (si *SpanIterator) Err() error { | |
| - if si.err != nil { | |
| - return si.err | |
| - } | |
| - return si.backing.err() | |
| -} | |
| - | |
| -// Next retrieves the next span in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into span, | |
| -// and false if there are no more elements or if an error was encountered. When | |
| -// Next returns false, the user should call the Err method to verify the | |
| -// existence of an error. | |
| -func (si *SpanIterator) Next(span *roachpb.Span) bool { | |
| - wrapper := resultWrapper{} | |
| - | |
| - for si.backing.next(&wrapper) { | |
| - if si.filter == nil || si.filter(wrapper.key) { | |
| - sp, err := decodeSpanSSTKey(wrapper.key.Key) | |
| - if err != nil { | |
| - si.err = err | |
| - return false | |
| - } | |
| - | |
| - *span = sp | |
| - return true | |
| - } | |
| - } | |
| - | |
| - return false | |
| -} | |
| - | |
| -// FileIterator is a simple iterator to iterate over stats.TableStatisticProtos. | |
| -type FileIterator struct { | |
| - mergedIterator storage.SimpleMVCCIterator | |
| - backingIterators []storage.SimpleMVCCIterator | |
| - err error | |
| -} | |
| - | |
| -// FileIter creates a new FileIterator for the backup metadata. | |
| -func (b *BackupMetadata) FileIter(ctx context.Context) FileIterator { | |
| - fileInfoIter := makeBytesIter(ctx, b.store, b.filename, []byte(sstFilesPrefix), b.enc, false) | |
| - defer fileInfoIter.close() | |
| - | |
| - var iters []storage.SimpleMVCCIterator | |
| - var encOpts *roachpb.FileEncryptionOptions | |
| - if b.enc != nil { | |
| - key, err := getEncryptionKey(ctx, b.enc, b.store.Settings(), b.store.ExternalIOConf()) | |
| - if err != nil { | |
| - return FileIterator{err: err} | |
| - } | |
| - encOpts = &roachpb.FileEncryptionOptions{Key: key} | |
| - } | |
| - | |
| - result := resultWrapper{} | |
| - for fileInfoIter.next(&result) { | |
| - path, err := decodeUnsafeFileInfoSSTKey(result.key.Key) | |
| - if err != nil { | |
| - break | |
| - } | |
| - | |
| - iter, err := storageccl.ExternalSSTReader(ctx, b.store, path, encOpts) | |
| - if err != nil { | |
| - return FileIterator{err: err} | |
| - } | |
| - iters = append(iters, iter) | |
| - } | |
| - | |
| - if fileInfoIter.err() != nil { | |
| - return FileIterator{err: fileInfoIter.err()} | |
| - } | |
| - | |
| - mergedIter := storage.MakeMultiIterator(iters) | |
| - mergedIter.SeekGE(storage.MVCCKey{}) | |
| - return FileIterator{mergedIterator: mergedIter, backingIterators: iters} | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (fi *FileIterator) Close() { | |
| - for _, it := range fi.backingIterators { | |
| - it.Close() | |
| - } | |
| - fi.mergedIterator = nil | |
| - fi.backingIterators = fi.backingIterators[:0] | |
| -} | |
| - | |
| -// Err returns the iterator's error. | |
| -func (fi *FileIterator) Err() error { | |
| - return fi.err | |
| -} | |
| - | |
| -// Next retrieves the next file in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into file, | |
| -// and false if there are no more elements or if an error was encountered. When | |
| -// Next returns false, the user should call the Err method to verify the | |
| -// existence of an error. | |
| -func (fi *FileIterator) Next(file *BackupManifest_File) bool { | |
| - if fi.err != nil { | |
| - return false | |
| - } | |
| - | |
| - valid, err := fi.mergedIterator.Valid() | |
| - if err != nil || !valid { | |
| - fi.err = err | |
| - return false | |
| - } | |
| - err = protoutil.Unmarshal(fi.mergedIterator.UnsafeValue(), file) | |
| - if err != nil { | |
| - fi.err = err | |
| - return false | |
| - } | |
| - | |
| - fi.mergedIterator.Next() | |
| - return true | |
| -} | |
| - | |
| -// DescIterator is a simple iterator to iterate over descpb.Descriptors. | |
| -type DescIterator struct { | |
| - backing bytesIter | |
| - err error | |
| -} | |
| - | |
| -// DescIter creates a new DescIterator for the backup metadata. | |
| -func (b *BackupMetadata) DescIter(ctx context.Context) DescIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstDescsPrefix), b.enc, true) | |
| - return DescIterator{ | |
| - backing: backing, | |
| - } | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (di *DescIterator) Close() { | |
| - di.backing.close() | |
| -} | |
| - | |
| -// Err returns the iterator's error. | |
| -func (di *DescIterator) Err() error { | |
| - if di.err != nil { | |
| - return di.err | |
| - } | |
| - return di.backing.err() | |
| -} | |
| - | |
| -// Next retrieves the next descriptor in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into desc , | |
| -// and false if there are no more elements or if an error was encountered. When | |
| -// Next returns false, the user should call the Err method to verify the | |
| -// existence of an error. | |
| -func (di *DescIterator) Next(desc *descpb.Descriptor) bool { | |
| - wrapper := resultWrapper{} | |
| - | |
| - for di.backing.next(&wrapper) { | |
| - err := protoutil.Unmarshal(wrapper.value, desc) | |
| - if err != nil { | |
| - di.err = err | |
| - return false | |
| - } | |
| - | |
| - tbl, db, typ, sc := descpb.FromDescriptor(desc) | |
| - if tbl != nil || db != nil || typ != nil || sc != nil { | |
| - return true | |
| - } | |
| - } | |
| - | |
| - return false | |
| -} | |
| - | |
| -// TenantIterator is a simple iterator to iterate over TenantInfoWithUsages. | |
| -type TenantIterator struct { | |
| - backing bytesIter | |
| - err error | |
| -} | |
| - | |
| -// TenantIter creates a new TenantIterator for the backup metadata. | |
| -func (b *BackupMetadata) TenantIter(ctx context.Context) TenantIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstTenantsPrefix), b.enc, false) | |
| - return TenantIterator{ | |
| - backing: backing, | |
| - } | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (ti *TenantIterator) Close() { | |
| - ti.backing.close() | |
| -} | |
| - | |
| -// Err returns the iterator's error. | |
| -func (ti *TenantIterator) Err() error { | |
| - if ti.err != nil { | |
| - return ti.err | |
| - } | |
| - return ti.backing.err() | |
| -} | |
| - | |
| -// Next retrieves the next tenant in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into tenant, | |
| -// and false if there are no more elements or if an error was encountered. When | |
| -// Next returns false, the user should call the Err method to verify the | |
| -// existence of an error. | |
| -func (ti *TenantIterator) Next(tenant *descpb.TenantInfoWithUsage) bool { | |
| - wrapper := resultWrapper{} | |
| - ok := ti.backing.next(&wrapper) | |
| - if !ok { | |
| - return false | |
| - } | |
| - | |
| - err := protoutil.Unmarshal(wrapper.value, tenant) | |
| - if err != nil { | |
| - ti.err = err | |
| - return false | |
| - } | |
| - | |
| - return true | |
| -} | |
| - | |
| -// DescriptorRevisionIterator is a simple iterator to iterate over BackupManifest_DescriptorRevisions. | |
| -type DescriptorRevisionIterator struct { | |
| - backing bytesIter | |
| - err error | |
| -} | |
| - | |
| -// DescriptorChangesIter creates a new DescriptorChangesIterator for the backup metadata. | |
| -func (b *BackupMetadata) DescriptorChangesIter(ctx context.Context) DescriptorRevisionIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstDescsPrefix), b.enc, false) | |
| - return DescriptorRevisionIterator{ | |
| - backing: backing, | |
| - } | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (dri *DescriptorRevisionIterator) Close() { | |
| - dri.backing.close() | |
| -} | |
| - | |
| -// Err returns the iterator's error. | |
| -func (dri *DescriptorRevisionIterator) Err() error { | |
| - if dri.err != nil { | |
| - return dri.err | |
| - } | |
| - return dri.backing.err() | |
| -} | |
| - | |
| -// Next retrieves the next descriptor revision in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into | |
| -// revision, and false if there are no more elements or if an error was | |
| -// encountered. When Next returns false, the user should call the Err method to | |
| -// verify the existence of an error. | |
| -func (dri *DescriptorRevisionIterator) Next(revision *BackupManifest_DescriptorRevision) bool { | |
| - wrapper := resultWrapper{} | |
| - ok := dri.backing.next(&wrapper) | |
| - if !ok { | |
| - return false | |
| - } | |
| - | |
| - err := unmarshalWrapper(&wrapper, revision) | |
| - if err != nil { | |
| - dri.err = err | |
| - return false | |
| - } | |
| - | |
| - return true | |
| -} | |
| - | |
| -func unmarshalWrapper(wrapper *resultWrapper, rev *BackupManifest_DescriptorRevision) error { | |
| - var desc *descpb.Descriptor | |
| - if len(wrapper.value) > 0 { | |
| - desc = &descpb.Descriptor{} | |
| - err := protoutil.Unmarshal(wrapper.value, desc) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - } | |
| - | |
| - id, err := decodeDescSSTKey(wrapper.key.Key) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - | |
| - *rev = BackupManifest_DescriptorRevision{ | |
| - Desc: desc, | |
| - ID: id, | |
| - Time: wrapper.key.Timestamp, | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -// StatsIterator is a simple iterator to iterate over stats.TableStatisticProtos. | |
| -type StatsIterator struct { | |
| - backing bytesIter | |
| - err error | |
| -} | |
| - | |
| -// StatsIter creates a new StatsIterator for the backup metadata. | |
| -func (b *BackupMetadata) StatsIter(ctx context.Context) StatsIterator { | |
| - backing := makeBytesIter(ctx, b.store, b.filename, []byte(sstStatsPrefix), b.enc, false) | |
| - return StatsIterator{ | |
| - backing: backing, | |
| - } | |
| -} | |
| - | |
| -// Close closes the iterator. | |
| -func (si *StatsIterator) Close() { | |
| - si.backing.close() | |
| -} | |
| - | |
| -// Err returns the iterator's error. | |
| -func (si *StatsIterator) Err() error { | |
| - if si.err != nil { | |
| - return si.err | |
| - } | |
| - return si.backing.err() | |
| -} | |
| - | |
| -// Next retrieves the next stats proto in the iterator. | |
| -// | |
| -// Next returns true if next element was successfully unmarshalled into | |
| -// statsPtr, and false if there are no more elements or if an error was | |
| -// encountered. When Next returns false, the user should call the Err method to verify the | |
| -// existence of an error. | |
| -func (si *StatsIterator) Next(statsPtr **stats.TableStatisticProto) bool { | |
| - wrapper := resultWrapper{} | |
| - ok := si.backing.next(&wrapper) | |
| - | |
| - if !ok { | |
| - return false | |
| - } | |
| - | |
| - var s stats.TableStatisticProto | |
| - err := protoutil.Unmarshal(wrapper.value, &s) | |
| - if err != nil { | |
| - si.err = err | |
| - return false | |
| - } | |
| - | |
| - *statsPtr = &s | |
| - return true | |
| -} | |
| - | |
| -type bytesIter struct { | |
| - Iter storage.SimpleMVCCIterator | |
| - | |
| - prefix []byte | |
| - useMVCCNext bool | |
| - iterError error | |
| -} | |
| - | |
| -func makeBytesIter( | |
| - ctx context.Context, | |
| - store cloud.ExternalStorage, | |
| - path string, | |
| - prefix []byte, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - useMVCCNext bool, | |
| -) bytesIter { | |
| - var encOpts *roachpb.FileEncryptionOptions | |
| - if enc != nil { | |
| - key, err := getEncryptionKey(ctx, enc, store.Settings(), store.ExternalIOConf()) | |
| - if err != nil { | |
| - return bytesIter{iterError: err} | |
| - } | |
| - encOpts = &roachpb.FileEncryptionOptions{Key: key} | |
| - } | |
| - | |
| - iter, err := storageccl.ExternalSSTReader(ctx, store, path, encOpts) | |
| - if err != nil { | |
| - return bytesIter{iterError: err} | |
| - } | |
| - | |
| - iter.SeekGE(storage.MakeMVCCMetadataKey(prefix)) | |
| - return bytesIter{ | |
| - Iter: iter, | |
| - prefix: prefix, | |
| - useMVCCNext: useMVCCNext, | |
| - } | |
| -} | |
| - | |
| -func (bi *bytesIter) next(resWrapper *resultWrapper) bool { | |
| - if bi.iterError != nil { | |
| - return false | |
| - } | |
| - | |
| - valid, err := bi.Iter.Valid() | |
| - if err != nil || !valid || !bytes.HasPrefix(bi.Iter.UnsafeKey().Key, bi.prefix) { | |
| - bi.close() | |
| - bi.iterError = err | |
| - return false | |
| - } | |
| - | |
| - key := bi.Iter.UnsafeKey() | |
| - resWrapper.key.Key = key.Key.Clone() | |
| - resWrapper.key.Timestamp = key.Timestamp | |
| - resWrapper.value = resWrapper.value[:0] | |
| - resWrapper.value = append(resWrapper.value, bi.Iter.UnsafeValue()...) | |
| - | |
| - if bi.useMVCCNext { | |
| - bi.Iter.NextKey() | |
| - } else { | |
| - bi.Iter.Next() | |
| - } | |
| - return true | |
| -} | |
| - | |
| -func (bi *bytesIter) err() error { | |
| - return bi.iterError | |
| -} | |
| - | |
| -func (bi *bytesIter) close() { | |
| - if bi.Iter != nil { | |
| - bi.Iter.Close() | |
| - bi.Iter = nil | |
| - } | |
| -} | |
| - | |
| -type resultWrapper struct { | |
| - key storage.MVCCKey | |
| - value []byte | |
| -} | |
| diff --git a/pkg/ccl/backupccl/backup_metadata_test.go b/pkg/ccl/backupccl/backup_metadata_test.go | |
| deleted file mode 100644 | |
| index bcfa56c9f6..0000000000 | |
| --- a/pkg/ccl/backupccl/backup_metadata_test.go | |
| +++ /dev/null | |
| @@ -1,281 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Licensed as a CockroachDB Enterprise file under the Cockroach Community | |
| -// License (the "License"); you may not use this file except in compliance with | |
| -// the License. You may obtain a copy of the License at | |
| -// | |
| -// https://github.com/cockroachdb/cockroach/blob/master/licenses/CCL.txt | |
| - | |
| -package backupccl | |
| - | |
| -import ( | |
| - "context" | |
| - "sort" | |
| - "testing" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/base" | |
| - "github.com/cockroachdb/cockroach/pkg/blobs" | |
| - "github.com/cockroachdb/cockroach/pkg/cloud" | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/security" | |
| - "github.com/cockroachdb/cockroach/pkg/sql" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/catalog/descpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/stats" | |
| - "github.com/cockroachdb/cockroach/pkg/testutils/testcluster" | |
| - "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| - "github.com/cockroachdb/cockroach/pkg/util/ioctx" | |
| - "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| - "github.com/cockroachdb/cockroach/pkg/util/log" | |
| - "github.com/cockroachdb/cockroach/pkg/util/protoutil" | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -func TestMetadataSST(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - ctx := context.Background() | |
| - const numAccounts = 1 | |
| - userfile := "userfile:///0" | |
| - tc, sqlDB, _, cleanupFn := backupRestoreTestSetup(t, singleNode, numAccounts, InitManualReplication) | |
| - defer cleanupFn() | |
| - | |
| - // Check that backup metadata is correct on full cluster backup. | |
| - sqlDB.Exec(t, `BACKUP TO $1`, userfile) | |
| - checkMetadata(ctx, t, tc, userfile) | |
| - | |
| - // Check for correct backup metadata on incremental backup with revision | |
| - // history. | |
| - sqlDB.Exec(t, `CREATE TABLE data.foo(k INT, v INT)`) | |
| - sqlDB.Exec(t, `CREATE INDEX idx ON data.bank (balance)`) | |
| - | |
| - sqlDB.Exec(t, `BACKUP TO $1 WITH revision_history`, userfile) | |
| - checkMetadata(ctx, t, tc, userfile) | |
| - | |
| - // Check for correct backup metadata on single table backups. | |
| - userfile1 := "userfile:///1" | |
| - sqlDB.Exec(t, `BACKUP TABLE data.bank TO $1 WITH revision_history`, userfile1) | |
| - checkMetadata(ctx, t, tc, userfile1) | |
| - | |
| - // Check for correct backup metadata on tenant backups. | |
| - userfile2 := "userfile:///2" | |
| - _, err := tc.Servers[0].StartTenant(ctx, base.TestTenantArgs{TenantID: roachpb.MakeTenantID(10)}) | |
| - require.NoError(t, err) | |
| - sqlDB.Exec(t, `BACKUP TENANT 10 TO $1`, userfile2) | |
| - checkMetadata(ctx, t, tc, userfile2) | |
| -} | |
| - | |
| -func checkMetadata( | |
| - ctx context.Context, t *testing.T, tc *testcluster.TestCluster, backupLoc string, | |
| -) { | |
| - store, err := cloud.ExternalStorageFromURI( | |
| - ctx, | |
| - backupLoc, | |
| - base.ExternalIODirConfig{}, | |
| - tc.Servers[0].ClusterSettings(), | |
| - blobs.TestEmptyBlobClientFactory, | |
| - security.RootUserName(), | |
| - tc.Servers[0].InternalExecutor().(*sql.InternalExecutor), tc.Servers[0].DB()) | |
| - if err != nil { | |
| - t.Fatal(err) | |
| - } | |
| - m, err := testingReadBackupManifest(ctx, store, backupManifestName) | |
| - if err != nil { | |
| - t.Fatal(err) | |
| - } | |
| - | |
| - bm, err := newBackupMetadata(ctx, store, metadataSSTName, nil) | |
| - if err != nil { | |
| - t.Fatal(err) | |
| - } | |
| - | |
| - checkManifest(t, m, bm) | |
| - // If there are descriptor changes, we only check those as they should have | |
| - // all changes as well as existing descriptors | |
| - if len(m.DescriptorChanges) > 0 { | |
| - checkDescriptorChanges(ctx, t, m, bm) | |
| - } else { | |
| - checkDescriptors(ctx, t, m, bm) | |
| - } | |
| - | |
| - checkSpans(ctx, t, m, bm) | |
| - // Don't check introduced spans on the first backup. | |
| - if m.StartTime != (hlc.Timestamp{}) { | |
| - checkIntroducedSpans(ctx, t, m, bm) | |
| - } | |
| - checkFiles(ctx, t, m, bm) | |
| - checkTenants(ctx, t, m, bm) | |
| - checkStats(ctx, t, store, m, bm) | |
| -} | |
| - | |
| -func checkManifest(t *testing.T, m *BackupManifest, bm *BackupMetadata) { | |
| - expectedManifest := *m | |
| - expectedManifest.Descriptors = nil | |
| - expectedManifest.DescriptorChanges = nil | |
| - expectedManifest.Files = nil | |
| - expectedManifest.Spans = nil | |
| - expectedManifest.IntroducedSpans = nil | |
| - expectedManifest.StatisticsFilenames = nil | |
| - expectedManifest.Tenants = nil | |
| - | |
| - require.Equal(t, expectedManifest, bm.BackupManifest) | |
| -} | |
| - | |
| -func checkDescriptors(ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata) { | |
| - var metaDescs []descpb.Descriptor | |
| - var desc descpb.Descriptor | |
| - | |
| - it := bm.DescIter(ctx) | |
| - defer it.Close() | |
| - for it.Next(&desc) { | |
| - metaDescs = append(metaDescs, desc) | |
| - } | |
| - | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - require.Equal(t, m.Descriptors, metaDescs) | |
| -} | |
| - | |
| -func checkDescriptorChanges( | |
| - ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata, | |
| -) { | |
| - var metaRevs []BackupManifest_DescriptorRevision | |
| - var rev BackupManifest_DescriptorRevision | |
| - it := bm.DescriptorChangesIter(ctx) | |
| - defer it.Close() | |
| - | |
| - for it.Next(&rev) { | |
| - metaRevs = append(metaRevs, rev) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - // Descriptor Changes are sorted by time in the manifest. | |
| - sort.Slice(metaRevs, func(i, j int) bool { | |
| - return metaRevs[i].Time.Less(metaRevs[j].Time) | |
| - }) | |
| - | |
| - require.Equal(t, m.DescriptorChanges, metaRevs) | |
| -} | |
| - | |
| -func checkFiles(ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata) { | |
| - var metaFiles []BackupManifest_File | |
| - var file BackupManifest_File | |
| - it := bm.FileIter(ctx) | |
| - defer it.Close() | |
| - | |
| - for it.Next(&file) { | |
| - metaFiles = append(metaFiles, file) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - require.Equal(t, m.Files, metaFiles) | |
| -} | |
| - | |
| -func checkSpans(ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata) { | |
| - var metaSpans []roachpb.Span | |
| - var span roachpb.Span | |
| - it := bm.SpanIter(ctx) | |
| - defer it.Close() | |
| - | |
| - for it.Next(&span) { | |
| - metaSpans = append(metaSpans, span) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - require.Equal(t, m.Spans, metaSpans) | |
| -} | |
| - | |
| -func checkIntroducedSpans( | |
| - ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata, | |
| -) { | |
| - var metaSpans []roachpb.Span | |
| - var span roachpb.Span | |
| - it := bm.IntroducedSpanIter(ctx) | |
| - defer it.Close() | |
| - for it.Next(&span) { | |
| - metaSpans = append(metaSpans, span) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - require.Equal(t, m.IntroducedSpans, metaSpans) | |
| -} | |
| - | |
| -func checkTenants(ctx context.Context, t *testing.T, m *BackupManifest, bm *BackupMetadata) { | |
| - var metaTenants []descpb.TenantInfoWithUsage | |
| - var tenant descpb.TenantInfoWithUsage | |
| - it := bm.TenantIter(ctx) | |
| - defer it.Close() | |
| - | |
| - for it.Next(&tenant) { | |
| - metaTenants = append(metaTenants, tenant) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - | |
| - require.Equal(t, m.Tenants, metaTenants) | |
| -} | |
| - | |
| -func checkStats( | |
| - ctx context.Context, | |
| - t *testing.T, | |
| - store cloud.ExternalStorage, | |
| - m *BackupManifest, | |
| - bm *BackupMetadata, | |
| -) { | |
| - expectedStats, err := getStatisticsFromBackup(ctx, store, nil, *m) | |
| - if err != nil { | |
| - t.Fatal(err) | |
| - } | |
| - | |
| - var metaStats = make([]*stats.TableStatisticProto, 0) | |
| - var s *stats.TableStatisticProto | |
| - it := bm.StatsIter(ctx) | |
| - defer it.Close() | |
| - | |
| - for it.Next(&s) { | |
| - metaStats = append(metaStats, s) | |
| - } | |
| - if it.Err() != nil { | |
| - t.Fatal(it.Err()) | |
| - } | |
| - require.Equal(t, expectedStats, metaStats) | |
| -} | |
| - | |
| -func testingReadBackupManifest( | |
| - ctx context.Context, store cloud.ExternalStorage, file string, | |
| -) (*BackupManifest, error) { | |
| - r, err := store.ReadFile(ctx, file) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - defer r.Close(ctx) | |
| - | |
| - bytes, err := ioctx.ReadAll(ctx, r) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - if isGZipped(bytes) { | |
| - descBytes, err := decompressData(ctx, nil, bytes) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - bytes = descBytes | |
| - } | |
| - | |
| - var m BackupManifest | |
| - if err := protoutil.Unmarshal(bytes, &m); err != nil { | |
| - return nil, err | |
| - } | |
| - return &m, nil | |
| -} | |
| diff --git a/pkg/ccl/backupccl/backup_planning.go b/pkg/ccl/backupccl/backup_planning.go | |
| index e0b642c2a4..d618e186f5 100644 | |
| --- a/pkg/ccl/backupccl/backup_planning.go | |
| +++ b/pkg/ccl/backupccl/backup_planning.go | |
| @@ -58,16 +58,15 @@ import ( | |
| ) | |
| const ( | |
| - backupOptRevisionHistory = "revision_history" | |
| - backupOptEncPassphrase = "encryption_passphrase" | |
| - backupOptEncKMS = "kms" | |
| - backupOptWithPrivileges = "privileges" | |
| - backupOptAsJSON = "as_json" | |
| - backupOptWithDebugIDs = "debug_ids" | |
| - backupOptIncStorage = "incremental_location" | |
| - localityURLParam = "COCKROACH_LOCALITY" | |
| - defaultLocalityValue = "default" | |
| - backupOptDebugMetadataSST = "debug_dump_metadata_sst" | |
| + backupOptRevisionHistory = "revision_history" | |
| + backupOptEncPassphrase = "encryption_passphrase" | |
| + backupOptEncKMS = "kms" | |
| + backupOptWithPrivileges = "privileges" | |
| + backupOptAsJSON = "as_json" | |
| + backupOptWithDebugIDs = "debug_ids" | |
| + backupOptIncStorage = "incremental_location" | |
| + localityURLParam = "COCKROACH_LOCALITY" | |
| + defaultLocalityValue = "default" | |
| ) | |
| type tableAndIndex struct { | |
| @@ -515,7 +514,7 @@ func checkPrivilegesForBackup( | |
| "only users with the admin role are allowed to perform full cluster backups") | |
| } | |
| // Do not allow tenant backups. | |
| - if backupStmt.Targets != nil && backupStmt.Targets.TenantID.IsSet() { | |
| + if backupStmt.Targets != nil && backupStmt.Targets.Tenant != (roachpb.TenantID{}) { | |
| return pgerror.Newf( | |
| pgcode.InsufficientPrivilege, | |
| "only users with the admin role can perform BACKUP TENANT") | |
| @@ -782,11 +781,11 @@ func backupPlanHook( | |
| } | |
| } | |
| - if backupStmt.Targets != nil && backupStmt.Targets.TenantID.IsSet() { | |
| + if backupStmt.Targets != nil && backupStmt.Targets.Tenant != (roachpb.TenantID{}) { | |
| if !p.ExecCfg().Codec.ForSystemTenant() { | |
| return pgerror.Newf(pgcode.InsufficientPrivilege, "only the system tenant can backup other tenants") | |
| } | |
| - initialDetails.SpecificTenantIds = []roachpb.TenantID{backupStmt.Targets.TenantID.TenantID} | |
| + initialDetails.SpecificTenantIds = []roachpb.TenantID{backupStmt.Targets.Tenant} | |
| } | |
| jobID := p.ExecCfg().JobRegistry.MakeJobID() | |
| diff --git a/pkg/ccl/backupccl/backup_processor.go b/pkg/ccl/backupccl/backup_processor.go | |
| index 55db29c741..b7fc0a0e0e 100644 | |
| --- a/pkg/ccl/backupccl/backup_processor.go | |
| +++ b/pkg/ccl/backupccl/backup_processor.go | |
| @@ -44,7 +44,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/logtags" | |
| gogotypes "github.com/gogo/protobuf/types" | |
| "github.com/kr/pretty" | |
| ) | |
| @@ -178,7 +177,6 @@ func newBackupDataProcessor( | |
| // Start is part of the RowSource interface. | |
| func (bp *backupDataProcessor) Start(ctx context.Context) { | |
| - ctx = logtags.AddTag(ctx, "job", bp.spec.JobID) | |
| ctx = bp.StartInternal(ctx, backupProcessorName) | |
| ctx, cancel := context.WithCancel(ctx) | |
| bp.cancelAndWaitForWorker = func() { | |
| diff --git a/pkg/ccl/backupccl/backup_processor_planning.go b/pkg/ccl/backupccl/backup_processor_planning.go | |
| index bc62ad876a..97fa9f6e16 100644 | |
| --- a/pkg/ccl/backupccl/backup_processor_planning.go | |
| +++ b/pkg/ccl/backupccl/backup_processor_planning.go | |
| @@ -24,6 +24,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| "github.com/cockroachdb/errors" | |
| + "github.com/cockroachdb/logtags" | |
| ) | |
| func distBackupPlanSpecs( | |
| @@ -31,7 +32,6 @@ func distBackupPlanSpecs( | |
| planCtx *sql.PlanningCtx, | |
| execCtx sql.JobExecContext, | |
| dsp *sql.DistSQLPlanner, | |
| - jobID int64, | |
| spans roachpb.Spans, | |
| introducedSpans roachpb.Spans, | |
| pkIDs map[uint64]bool, | |
| @@ -92,7 +92,6 @@ func distBackupPlanSpecs( | |
| sqlInstanceIDToSpec := make(map[base.SQLInstanceID]*execinfrapb.BackupDataSpec) | |
| for _, partition := range spanPartitions { | |
| spec := &execinfrapb.BackupDataSpec{ | |
| - JobID: jobID, | |
| Spans: partition.Spans, | |
| DefaultURI: defaultURI, | |
| URIsByLocalityKV: urisByLocalityKV, | |
| @@ -114,7 +113,6 @@ func distBackupPlanSpecs( | |
| // which is not the leaseholder for any of the spans, but is for an | |
| // introduced span. | |
| spec := &execinfrapb.BackupDataSpec{ | |
| - JobID: jobID, | |
| IntroducedSpans: partition.Spans, | |
| DefaultURI: defaultURI, | |
| URIsByLocalityKV: urisByLocalityKV, | |
| @@ -155,6 +153,7 @@ func distBackup( | |
| ) error { | |
| ctx, span := tracing.ChildSpan(ctx, "backup-distsql") | |
| defer span.Finish() | |
| + ctx = logtags.AddTag(ctx, "backup-distsql", nil) | |
| evalCtx := execCtx.ExtendedEvalContext() | |
| var noTxn *kv.Txn | |
| diff --git a/pkg/ccl/backupccl/backup_test.go b/pkg/ccl/backupccl/backup_test.go | |
| index 767a78d3e1..57408ca3f3 100644 | |
| --- a/pkg/ccl/backupccl/backup_test.go | |
| +++ b/pkg/ccl/backupccl/backup_test.go | |
| @@ -91,7 +91,7 @@ import ( | |
| "github.com/cockroachdb/errors/oserror" | |
| "github.com/cockroachdb/logtags" | |
| "github.com/gogo/protobuf/proto" | |
| - pgx "github.com/jackc/pgx/v4" | |
| + "github.com/jackc/pgx/v4" | |
| "github.com/kr/pretty" | |
| "github.com/lib/pq" | |
| "github.com/stretchr/testify/assert" | |
| @@ -9353,8 +9353,6 @@ func TestGCDropIndexSpanExpansion(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| - skip.UnderRace(t, "takes >1 min under race") | |
| - | |
| aboutToGC := make(chan struct{}) | |
| allowGC := make(chan struct{}) | |
| var gcJobID jobspb.JobID | |
| diff --git a/pkg/ccl/backupccl/create_scheduled_backup.go b/pkg/ccl/backupccl/create_scheduled_backup.go | |
| index bccdcd5ba2..47dda37e09 100644 | |
| --- a/pkg/ccl/backupccl/create_scheduled_backup.go | |
| +++ b/pkg/ccl/backupccl/create_scheduled_backup.go | |
| @@ -20,6 +20,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/jobs" | |
| "github.com/cockroachdb/cockroach/pkg/jobs/jobspb" | |
| "github.com/cockroachdb/cockroach/pkg/kv" | |
| + "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/scheduledjobs" | |
| "github.com/cockroachdb/cockroach/pkg/security" | |
| "github.com/cockroachdb/cockroach/pkg/server/telemetry" | |
| @@ -269,7 +270,7 @@ func canChainProtectedTimestampRecords(p sql.PlanHookState, eval *scheduledBacku | |
| } | |
| // Return true if the backup has table targets or is backing up a tenant. | |
| - return eval.Targets.Tables != nil || eval.Targets.TenantID.IsSet() | |
| + return eval.Targets.Tables != nil || eval.Targets.Tenant != roachpb.TenantID{} | |
| } | |
| // doCreateBackupSchedule creates requested schedule (or schedules). | |
| diff --git a/pkg/ccl/backupccl/manifest_handling.go b/pkg/ccl/backupccl/manifest_handling.go | |
| index 4cc4f60f97..06a4d80ddc 100644 | |
| --- a/pkg/ccl/backupccl/manifest_handling.go | |
| +++ b/pkg/ccl/backupccl/manifest_handling.go | |
| @@ -27,7 +27,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/jobs/jobspb" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/security" | |
| - "github.com/cockroachdb/cockroach/pkg/settings" | |
| "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/descbuilder" | |
| @@ -99,13 +98,6 @@ const ( | |
| latestHistoryDirectory = "latest" | |
| ) | |
| -var writeMetadataSST = settings.RegisterBoolSetting( | |
| - settings.TenantWritable, | |
| - "kv.bulkio.write_metadata_sst.enabled", | |
| - "write experimental new format BACKUP metadata file", | |
| - true, | |
| -) | |
| - | |
| // isGZipped detects whether the given bytes represent GZipped data. This check | |
| // is used rather than a standard implementation such as http.DetectContentType | |
| // since some zipped data may be mis-identified by that method. We've seen | |
| diff --git a/pkg/ccl/backupccl/restore_data_processor.go b/pkg/ccl/backupccl/restore_data_processor.go | |
| index 987e5c0293..504532bf91 100644 | |
| --- a/pkg/ccl/backupccl/restore_data_processor.go | |
| +++ b/pkg/ccl/backupccl/restore_data_processor.go | |
| @@ -35,7 +35,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/protoutil" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/logtags" | |
| gogotypes "github.com/gogo/protobuf/types" | |
| ) | |
| @@ -55,6 +54,9 @@ type restoreDataProcessor struct { | |
| // is updated, the job should be PAUSEd and RESUMEd for the new setting to | |
| // take effect. | |
| numWorkers int | |
| + // flushBytes is the maximum buffer size used when creating SSTs to flush. It | |
| + // remains constant over the lifetime of the processor. | |
| + flushBytes int64 | |
| // phaseGroup manages the phases of the restore: | |
| // 1) reading entries from the input | |
| @@ -145,6 +147,7 @@ func newRestoreDataProcessor( | |
| progCh: make(chan RestoreProgress, maxConcurrentRestoreWorkers), | |
| metaCh: make(chan *execinfrapb.ProducerMetadata, 1), | |
| numWorkers: int(numRestoreWorkers.Get(sv)), | |
| + flushBytes: bulk.IngestFileSize(flowCtx.Cfg.Settings), | |
| } | |
| if err := rd.Init(rd, post, restoreDataOutputTypes, flowCtx, processorID, output, nil, /* memMonitor */ | |
| @@ -162,7 +165,6 @@ func newRestoreDataProcessor( | |
| // Start is part of the RowSource interface. | |
| func (rd *restoreDataProcessor) Start(ctx context.Context) { | |
| - ctx = logtags.AddTag(ctx, "job", rd.spec.JobID) | |
| ctx = rd.StartInternal(ctx, restoreDataProcName) | |
| rd.input.Start(ctx) | |
| @@ -427,6 +429,7 @@ func (rd *restoreDataProcessor) processRestoreSpanEntry( | |
| batcher, err := bulk.MakeSSTBatcher(ctx, | |
| db, | |
| evalCtx.Settings, | |
| + func() int64 { return rd.flushBytes }, | |
| disallowShadowingBelow, | |
| writeAtBatchTS, | |
| ) | |
| diff --git a/pkg/ccl/backupccl/restore_data_processor_test.go b/pkg/ccl/backupccl/restore_data_processor_test.go | |
| index 3031186993..7c31e027bb 100644 | |
| --- a/pkg/ccl/backupccl/restore_data_processor_test.go | |
| +++ b/pkg/ccl/backupccl/restore_data_processor_test.go | |
| @@ -221,7 +221,7 @@ func runTestIngest(t *testing.T, init func(*cluster.Settings)) { | |
| if r < 0 { | |
| return nil | |
| } | |
| - return roachpb.NewError(roachpb.NewAmbiguousResultErrorf("%d", r)) | |
| + return roachpb.NewError(roachpb.NewAmbiguousResultError(strconv.Itoa(int(r)))) | |
| }, | |
| }, | |
| }} | |
| diff --git a/pkg/ccl/backupccl/restore_job.go b/pkg/ccl/backupccl/restore_job.go | |
| index 22d53ddc36..3e9d9acb3a 100644 | |
| --- a/pkg/ccl/backupccl/restore_job.go | |
| +++ b/pkg/ccl/backupccl/restore_job.go | |
| @@ -357,7 +357,6 @@ func restore( | |
| return distRestore( | |
| ctx, | |
| execCtx, | |
| - int64(job.ID()), | |
| importSpanChunks, | |
| dataToRestore.getPKIDs(), | |
| encryption, | |
| diff --git a/pkg/ccl/backupccl/restore_planning.go b/pkg/ccl/backupccl/restore_planning.go | |
| index 42855b3579..b080898d59 100644 | |
| --- a/pkg/ccl/backupccl/restore_planning.go | |
| +++ b/pkg/ccl/backupccl/restore_planning.go | |
| @@ -868,9 +868,7 @@ func maybeUpgradeDescriptors(descs []catalog.Descriptor, skipFKsWithNoMatchingTa | |
| } else { | |
| b = desc.NewBuilder() | |
| } | |
| - if err := b.RunPostDeserializationChanges(); err != nil { | |
| - return errors.NewAssertionErrorWithWrappedErrf(err, "error during RunPostDeserializationChanges") | |
| - } | |
| + b.RunPostDeserializationChanges() | |
| err := b.RunRestoreChanges(func(id descpb.ID) catalog.Descriptor { | |
| for _, d := range descs { | |
| if d.GetID() == id { | |
| @@ -1111,7 +1109,7 @@ func restorePlanHook( | |
| var newTenantIDFn func() (*roachpb.TenantID, error) | |
| if restoreStmt.Options.AsTenant != nil { | |
| - if restoreStmt.DescriptorCoverage == tree.AllDescriptors || !restoreStmt.Targets.TenantID.IsSet() { | |
| + if restoreStmt.DescriptorCoverage == tree.AllDescriptors || !restoreStmt.Targets.Tenant.IsSet() { | |
| err := errors.Errorf("%q can only be used when running RESTORE TENANT for a single tenant", restoreOptAsTenant) | |
| return nil, nil, nil, false, err | |
| } | |
| @@ -1288,7 +1286,7 @@ func checkPrivilegesForRestore( | |
| "only users with the admin role are allowed to restore full cluster backups") | |
| } | |
| // Do not allow tenant restores. | |
| - if restoreStmt.Targets.TenantID.IsSet() { | |
| + if restoreStmt.Targets.Tenant != (roachpb.TenantID{}) { | |
| return pgerror.Newf( | |
| pgcode.InsufficientPrivilege, | |
| "only users with the admin role can perform RESTORE TENANT") | |
| diff --git a/pkg/ccl/backupccl/restore_processor_planning.go b/pkg/ccl/backupccl/restore_processor_planning.go | |
| index 04dda3069d..39525b357f 100644 | |
| --- a/pkg/ccl/backupccl/restore_processor_planning.go | |
| +++ b/pkg/ccl/backupccl/restore_processor_planning.go | |
| @@ -26,6 +26,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/sql/types" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/errors" | |
| + "github.com/cockroachdb/logtags" | |
| ) | |
| // distRestore plans a 2 stage distSQL flow for a distributed restore. It | |
| @@ -40,7 +41,6 @@ import ( | |
| func distRestore( | |
| ctx context.Context, | |
| execCtx sql.JobExecContext, | |
| - jobID int64, | |
| chunks [][]execinfrapb.RestoreSpanEntry, | |
| pkIDs map[uint64]bool, | |
| encryption *jobspb.BackupEncryptionOptions, | |
| @@ -49,6 +49,7 @@ func distRestore( | |
| restoreTime hlc.Timestamp, | |
| progCh chan *execinfrapb.RemoteProducerMetadata_BulkProcessorProgress, | |
| ) error { | |
| + ctx = logtags.AddTag(ctx, "restore-distsql", nil) | |
| defer close(progCh) | |
| var noTxn *kv.Txn | |
| @@ -88,7 +89,6 @@ func distRestore( | |
| } | |
| restoreDataSpec := execinfrapb.RestoreDataSpec{ | |
| - JobID: jobID, | |
| RestoreTime: restoreTime, | |
| Encryption: fileEncryption, | |
| TableRekeys: tableRekeys, | |
| diff --git a/pkg/ccl/backupccl/show.go b/pkg/ccl/backupccl/show.go | |
| index af16e14b0b..9e90e0865c 100644 | |
| --- a/pkg/ccl/backupccl/show.go | |
| +++ b/pkg/ccl/backupccl/show.go | |
| @@ -34,7 +34,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/sql/sem/tree" | |
| "github.com/cockroachdb/cockroach/pkg/sql/types" | |
| "github.com/cockroachdb/cockroach/pkg/util/encoding" | |
| - "github.com/cockroachdb/cockroach/pkg/util/json" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/mon" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| @@ -164,54 +163,6 @@ func (m manifestInfoReader) showBackup( | |
| return nil | |
| } | |
| -type metadataSSTInfoReader struct{} | |
| - | |
| -var _ backupInfoReader = manifestInfoReader{} | |
| - | |
| -func (m metadataSSTInfoReader) header() colinfo.ResultColumns { | |
| - return colinfo.ResultColumns{ | |
| - {Name: "file", Typ: types.String}, | |
| - {Name: "key", Typ: types.String}, | |
| - {Name: "detail", Typ: types.Jsonb}, | |
| - } | |
| -} | |
| - | |
| -func (m metadataSSTInfoReader) showBackup( | |
| - ctx context.Context, | |
| - mem *mon.BoundAccount, | |
| - store cloud.ExternalStorage, | |
| - incStore cloud.ExternalStorage, | |
| - enc *jobspb.BackupEncryptionOptions, | |
| - incPaths []string, | |
| - resultsCh chan<- tree.Datums, | |
| -) error { | |
| - filename := metadataSSTName | |
| - push := func(_, readable string, value json.JSON) error { | |
| - val := tree.DNull | |
| - if value != nil { | |
| - val = tree.NewDJSON(value) | |
| - } | |
| - select { | |
| - case <-ctx.Done(): | |
| - return ctx.Err() | |
| - case resultsCh <- []tree.Datum{tree.NewDString(filename), tree.NewDString(readable), val}: | |
| - return nil | |
| - } | |
| - } | |
| - | |
| - if err := DebugDumpMetadataSST(ctx, store, filename, enc, push); err != nil { | |
| - return err | |
| - } | |
| - | |
| - for _, i := range incPaths { | |
| - filename = strings.TrimSuffix(i, backupManifestName) + metadataSSTName | |
| - if err := DebugDumpMetadataSST(ctx, incStore, filename, enc, push); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| - | |
| // showBackupPlanHook implements PlanHookFn. | |
| func showBackupPlanHook( | |
| ctx context.Context, stmt tree.Statement, p sql.PlanHookState, | |
| @@ -239,13 +190,12 @@ func showBackupPlanHook( | |
| } | |
| expected := map[string]sql.KVStringOptValidate{ | |
| - backupOptEncPassphrase: sql.KVStringOptRequireValue, | |
| - backupOptEncKMS: sql.KVStringOptRequireValue, | |
| - backupOptWithPrivileges: sql.KVStringOptRequireNoValue, | |
| - backupOptAsJSON: sql.KVStringOptRequireNoValue, | |
| - backupOptWithDebugIDs: sql.KVStringOptRequireNoValue, | |
| - backupOptIncStorage: sql.KVStringOptRequireValue, | |
| - backupOptDebugMetadataSST: sql.KVStringOptRequireNoValue, | |
| + backupOptEncPassphrase: sql.KVStringOptRequireValue, | |
| + backupOptEncKMS: sql.KVStringOptRequireValue, | |
| + backupOptWithPrivileges: sql.KVStringOptRequireNoValue, | |
| + backupOptAsJSON: sql.KVStringOptRequireNoValue, | |
| + backupOptWithDebugIDs: sql.KVStringOptRequireNoValue, | |
| + backupOptIncStorage: sql.KVStringOptRequireValue, | |
| } | |
| optsFn, err := p.TypeAsStringOpts(ctx, backup.Options, expected) | |
| if err != nil { | |
| @@ -261,22 +211,18 @@ func showBackupPlanHook( | |
| } | |
| var infoReader backupInfoReader | |
| - if _, dumpSST := opts[backupOptDebugMetadataSST]; dumpSST { | |
| - infoReader = metadataSSTInfoReader{} | |
| - } else { | |
| - var shower backupShower | |
| - switch backup.Details { | |
| - case tree.BackupRangeDetails: | |
| - shower = backupShowerRanges | |
| - case tree.BackupFileDetails: | |
| - shower = backupShowerFiles | |
| - case tree.BackupManifestAsJSON: | |
| - shower = jsonShower | |
| - default: | |
| - shower = backupShowerDefault(ctx, p, backup.ShouldIncludeSchemas, opts) | |
| - } | |
| - infoReader = manifestInfoReader{shower} | |
| + var shower backupShower | |
| + switch backup.Details { | |
| + case tree.BackupRangeDetails: | |
| + shower = backupShowerRanges | |
| + case tree.BackupFileDetails: | |
| + shower = backupShowerFiles | |
| + case tree.BackupManifestAsJSON: | |
| + shower = jsonShower | |
| + default: | |
| + shower = backupShowerDefault(ctx, p, backup.ShouldIncludeSchemas, opts) | |
| } | |
| + infoReader = manifestInfoReader{shower} | |
| fn := func(ctx context.Context, _ []sql.PlanNode, resultsCh chan<- tree.Datums) error { | |
| // TODO(dan): Move this span into sql. | |
| diff --git a/pkg/ccl/backupccl/split_and_scatter_processor.go b/pkg/ccl/backupccl/split_and_scatter_processor.go | |
| index a8cf6daa27..b458448535 100644 | |
| --- a/pkg/ccl/backupccl/split_and_scatter_processor.go | |
| +++ b/pkg/ccl/backupccl/split_and_scatter_processor.go | |
| @@ -29,7 +29,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/protoutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/logtags" | |
| ) | |
| type splitAndScatterer interface { | |
| @@ -245,7 +244,6 @@ func newSplitAndScatterProcessor( | |
| // Start is part of the RowSource interface. | |
| func (ssp *splitAndScatterProcessor) Start(ctx context.Context) { | |
| - ctx = logtags.AddTag(ctx, "job", ssp.spec.JobID) | |
| ctx = ssp.StartInternal(ctx, splitAndScatterProcessorName) | |
| // Note that the loop over doneScatterCh in Next should prevent the goroutine | |
| // below from leaking when there are no errors. However, if that loop needs to | |
| diff --git a/pkg/ccl/backupccl/targets.go b/pkg/ccl/backupccl/targets.go | |
| index 6ebfec397f..b5593ad14e 100644 | |
| --- a/pkg/ccl/backupccl/targets.go | |
| +++ b/pkg/ccl/backupccl/targets.go | |
| @@ -366,15 +366,15 @@ func selectTargets( | |
| return systemTables, nil, nil, nil | |
| } | |
| - if targets.TenantID.IsSet() { | |
| + if targets.Tenant != (roachpb.TenantID{}) { | |
| for _, tenant := range lastBackupManifest.GetTenants() { | |
| // TODO(dt): for now it is zero-or-one but when that changes, we should | |
| // either keep it sorted or build a set here. | |
| - if tenant.ID == targets.TenantID.ToUint64() { | |
| + if tenant.ID == targets.Tenant.ToUint64() { | |
| return nil, nil, []descpb.TenantInfoWithUsage{tenant}, nil | |
| } | |
| } | |
| - return nil, nil, nil, errors.Errorf("tenant %d not in backup", targets.TenantID.ToUint64()) | |
| + return nil, nil, nil, errors.Errorf("tenant %d not in backup", targets.Tenant.ToUint64()) | |
| } | |
| matched, err := backupresolver.DescriptorsMatchingTargets(ctx, | |
| diff --git a/pkg/ccl/benchccl/rttanalysisccl/BUILD.bazel b/pkg/ccl/benchccl/rttanalysisccl/BUILD.bazel | |
| index d502e43257..a7a0b17dc9 100644 | |
| --- a/pkg/ccl/benchccl/rttanalysisccl/BUILD.bazel | |
| +++ b/pkg/ccl/benchccl/rttanalysisccl/BUILD.bazel | |
| @@ -8,7 +8,6 @@ go_test( | |
| "multi_region_bench_test.go", | |
| ], | |
| data = glob(["testdata/**"]), | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/bench/rttanalysis", | |
| diff --git a/pkg/ccl/changefeedccl/BUILD.bazel b/pkg/ccl/changefeedccl/BUILD.bazel | |
| index 43da7d3ba7..9f20d86960 100644 | |
| --- a/pkg/ccl/changefeedccl/BUILD.bazel | |
| +++ b/pkg/ccl/changefeedccl/BUILD.bazel | |
| @@ -108,7 +108,6 @@ go_library( | |
| "@com_github_cockroachdb_apd_v3//:apd", | |
| "@com_github_cockroachdb_errors//:errors", | |
| "@com_github_cockroachdb_logtags//:logtags", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| "@com_github_google_btree//:btree", | |
| "@com_github_linkedin_goavro_v2//:goavro", | |
| "@com_github_shopify_sarama//:sarama", | |
| @@ -182,6 +181,7 @@ go_test( | |
| "//pkg/server/status", | |
| "//pkg/server/telemetry", | |
| "//pkg/settings/cluster", | |
| + "//pkg/sql", | |
| "//pkg/sql/catalog", | |
| "//pkg/sql/catalog/bootstrap", | |
| "//pkg/sql/catalog/descbuilder", | |
| diff --git a/pkg/ccl/changefeedccl/avro.go b/pkg/ccl/changefeedccl/avro.go | |
| index 85d48b002f..87dcabe960 100644 | |
| --- a/pkg/ccl/changefeedccl/avro.go | |
| +++ b/pkg/ccl/changefeedccl/avro.go | |
| @@ -18,7 +18,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/geo" | |
| "github.com/cockroachdb/cockroach/pkg/geo/geopb" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/catalog/descpb" | |
| "github.com/cockroachdb/cockroach/pkg/sql/rowenc" | |
| "github.com/cockroachdb/cockroach/pkg/sql/sem/tree" | |
| "github.com/cockroachdb/cockroach/pkg/sql/types" | |
| @@ -757,21 +756,11 @@ const ( | |
| // appended to the end of the avro record's name. | |
| func tableToAvroSchema( | |
| tableDesc catalog.TableDescriptor, | |
| - familyID descpb.FamilyID, | |
| nameSuffix string, | |
| namespace string, | |
| virtualColumnVisibility string, | |
| ) (*avroDataRecord, error) { | |
| - family, err := tableDesc.FindFamilyByID(familyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - var name string | |
| - if tableDesc.NumFamilies() > 1 { | |
| - name = SQLNameToAvroName(tableDesc.GetName() + family.Name) | |
| - } else { | |
| - name = SQLNameToAvroName(tableDesc.GetName()) | |
| - } | |
| + name := SQLNameToAvroName(tableDesc.GetName()) | |
| if nameSuffix != avroSchemaNoSuffix { | |
| name = name + `_` + nameSuffix | |
| } | |
| @@ -785,29 +774,18 @@ func tableToAvroSchema( | |
| colIdxByFieldIdx: make(map[int]int), | |
| fieldIdxByColIdx: make(map[int]int), | |
| } | |
| - | |
| - include := make(map[descpb.ColumnID]struct{}, len(family.ColumnIDs)) | |
| - var yes struct{} | |
| - for _, colID := range family.ColumnIDs { | |
| - include[colID] = yes | |
| - } | |
| - | |
| for _, col := range tableDesc.PublicColumns() { | |
| + if col.IsVirtual() && virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsOmitted) { | |
| + continue | |
| + } | |
| + field, err := columnToAvroSchema(col) | |
| if err != nil { | |
| return nil, err | |
| } | |
| - _, inFamily := include[col.GetID()] | |
| - virtual := col.IsVirtual() && virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsNull) | |
| - if inFamily || virtual { | |
| - field, err := columnToAvroSchema(col) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - schema.colIdxByFieldIdx[len(schema.Fields)] = col.Ordinal() | |
| - schema.fieldIdxByName[field.Name] = len(schema.Fields) | |
| - schema.fieldIdxByColIdx[col.Ordinal()] = len(schema.Fields) | |
| - schema.Fields = append(schema.Fields, field) | |
| - } | |
| + schema.colIdxByFieldIdx[len(schema.Fields)] = col.Ordinal() | |
| + schema.fieldIdxByName[field.Name] = len(schema.Fields) | |
| + schema.fieldIdxByColIdx[col.Ordinal()] = len(schema.Fields) | |
| + schema.Fields = append(schema.Fields, field) | |
| } | |
| schemaJSON, err := json.Marshal(schema) | |
| if err != nil { | |
| diff --git a/pkg/ccl/changefeedccl/avro_test.go b/pkg/ccl/changefeedccl/avro_test.go | |
| index 233f8acb41..296672626b 100644 | |
| --- a/pkg/ccl/changefeedccl/avro_test.go | |
| +++ b/pkg/ccl/changefeedccl/avro_test.go | |
| @@ -52,8 +52,6 @@ import ( | |
| var testTypes = make(map[string]*types.T) | |
| var testTypeResolver = tree.MakeTestingMapTypeResolver(testTypes) | |
| -const primary = descpb.FamilyID(0) | |
| - | |
| func makeTestSemaCtx() tree.SemaContext { | |
| testSemaCtx := tree.MakeSemaContext() | |
| testSemaCtx.TypeResolver = testTypeResolver | |
| @@ -79,13 +77,6 @@ func parseTableDesc(createTableStmt string) (catalog.TableDescriptor, error) { | |
| if err != nil { | |
| return nil, err | |
| } | |
| - columnNames := make([]string, len(mutDesc.PublicColumns())) | |
| - for i, col := range mutDesc.PublicColumns() { | |
| - columnNames[i] = col.GetName() | |
| - } | |
| - mutDesc.Families = []descpb.ColumnFamilyDescriptor{ | |
| - {ID: primary, Name: "primary", ColumnIDs: mutDesc.PublicColumnIDs(), ColumnNames: columnNames}, | |
| - } | |
| return mutDesc, descbuilder.ValidateSelf(mutDesc, clusterversion.TestingClusterVersion) | |
| } | |
| @@ -139,7 +130,7 @@ func parseAvroSchema(j string) (*avroDataRecord, error) { | |
| tableDesc := descpb.TableDescriptor{ | |
| Name: AvroNameToSQLName(s.Name), | |
| } | |
| - for i, f := range s.Fields { | |
| + for _, f := range s.Fields { | |
| // s.Fields[idx] has `Name` and `SchemaType` set but nothing else. | |
| // They're needed for serialization/deserialization, so fake out a | |
| // column descriptor so that we can reuse columnToAvroSchema to get | |
| @@ -148,19 +139,9 @@ func parseAvroSchema(j string) (*avroDataRecord, error) { | |
| if err != nil { | |
| return nil, err | |
| } | |
| - colDesc.ID = descpb.ColumnID(i) | |
| tableDesc.Columns = append(tableDesc.Columns, *colDesc) | |
| } | |
| - columnNames := make([]string, len(tableDesc.Columns)) | |
| - columnIDs := make([]descpb.ColumnID, len(tableDesc.Columns)) | |
| - for i, col := range tableDesc.Columns { | |
| - columnNames[i] = col.Name | |
| - columnIDs[i] = col.ID | |
| - } | |
| - tableDesc.Families = []descpb.ColumnFamilyDescriptor{ | |
| - {ID: primary, Name: "primary", ColumnIDs: columnIDs, ColumnNames: columnNames}, | |
| - } | |
| - return tableToAvroSchema(tabledesc.NewBuilder(&tableDesc).BuildImmutableTable(), primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + return tableToAvroSchema(tabledesc.NewBuilder(&tableDesc).BuildImmutableTable(), avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| } | |
| func avroFieldMetadataToColDesc(metadata string) (*descpb.ColumnDescriptor, error) { | |
| @@ -383,7 +364,7 @@ func TestAvroSchema(t *testing.T) { | |
| tableDesc, err := parseTableDesc( | |
| fmt.Sprintf(`CREATE TABLE "%s" %s`, test.name, test.schema)) | |
| require.NoError(t, err) | |
| - origSchema, err := tableToAvroSchema(tableDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + origSchema, err := tableToAvroSchema(tableDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| jsonSchema := origSchema.codec.Schema() | |
| roundtrippedSchema, err := parseAvroSchema(jsonSchema) | |
| @@ -419,7 +400,7 @@ func TestAvroSchema(t *testing.T) { | |
| t.Run("escaping", func(t *testing.T) { | |
| tableDesc, err := parseTableDesc(`CREATE TABLE "☃" (🍦 INT PRIMARY KEY)`) | |
| require.NoError(t, err) | |
| - tableSchema, err := tableToAvroSchema(tableDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + tableSchema, err := tableToAvroSchema(tableDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| require.Equal(t, | |
| `{"type":"record","name":"_u2603_","fields":[`+ | |
| @@ -626,7 +607,7 @@ func TestAvroSchema(t *testing.T) { | |
| rows, err := parseValues(tableDesc, `VALUES (1, `+test.sql+`)`) | |
| require.NoError(t, err) | |
| - schema, err := tableToAvroSchema(tableDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + schema, err := tableToAvroSchema(tableDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| textual, err := schema.textualFromRow(rows[0]) | |
| require.NoError(t, err) | |
| @@ -676,7 +657,7 @@ func TestAvroSchema(t *testing.T) { | |
| rows, err := parseValues(tableDesc, `VALUES (1, `+test.sql+`)`) | |
| require.NoError(t, err) | |
| - schema, err := tableToAvroSchema(tableDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + schema, err := tableToAvroSchema(tableDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| textual, err := schema.textualFromRow(rows[0]) | |
| require.NoError(t, err) | |
| @@ -779,12 +760,12 @@ func TestAvroMigration(t *testing.T) { | |
| writerDesc, err := parseTableDesc( | |
| fmt.Sprintf(`CREATE TABLE "%s" %s`, test.name, test.writerSchema)) | |
| require.NoError(t, err) | |
| - writerSchema, err := tableToAvroSchema(writerDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + writerSchema, err := tableToAvroSchema(writerDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| readerDesc, err := parseTableDesc( | |
| fmt.Sprintf(`CREATE TABLE "%s" %s`, test.name, test.readerSchema)) | |
| require.NoError(t, err) | |
| - readerSchema, err := tableToAvroSchema(readerDesc, primary, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + readerSchema, err := tableToAvroSchema(readerDesc, avroSchemaNoSuffix, "", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(t, err) | |
| writerRows, err := parseValues(writerDesc, `VALUES `+test.writerValues) | |
| @@ -861,7 +842,7 @@ func benchmarkEncodeType(b *testing.B, typ *types.T, encRow rowenc.EncDatumRow) | |
| tableDesc, err := parseTableDesc( | |
| fmt.Sprintf(`CREATE TABLE bench_table (bench_field %s)`, typ.SQLString())) | |
| require.NoError(b, err) | |
| - schema, err := tableToAvroSchema(tableDesc, primary, "suffix", "namespace", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| + schema, err := tableToAvroSchema(tableDesc, "suffix", "namespace", string(changefeedbase.OptVirtualColumnsOmitted)) | |
| require.NoError(b, err) | |
| b.ReportAllocs() | |
| diff --git a/pkg/ccl/changefeedccl/changefeed_processors.go b/pkg/ccl/changefeedccl/changefeed_processors.go | |
| index 62aeda2b82..74cab69d99 100644 | |
| --- a/pkg/ccl/changefeedccl/changefeed_processors.go | |
| +++ b/pkg/ccl/changefeedccl/changefeed_processors.go | |
| @@ -47,8 +47,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/uuid" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/logtags" | |
| - "github.com/cockroachdb/redact" | |
| ) | |
| type changeAggregator struct { | |
| @@ -204,9 +202,6 @@ func (ca *changeAggregator) MustBeStreaming() bool { | |
| // Start is part of the RowSource interface. | |
| func (ca *changeAggregator) Start(ctx context.Context) { | |
| - if ca.spec.JobID != 0 { | |
| - ctx = logtags.AddTag(ctx, "job", ca.spec.JobID) | |
| - } | |
| ctx = ca.StartInternal(ctx, changeAggregatorProcName) | |
| // Derive a separate context so that we can shutdown the poller. Note that | |
| @@ -356,30 +351,29 @@ func (ca *changeAggregator) makeKVFeedCfg( | |
| sf = schemafeed.DoNothingSchemaFeed | |
| } else { | |
| sf = schemafeed.New(ctx, cfg, schemaChangeEvents, AllTargets(ca.spec.Feed), | |
| - initialHighWater, &ca.metrics.SchemaFeedMetrics, ca.spec.Feed.Opts) | |
| + initialHighWater, &ca.metrics.SchemaFeedMetrics) | |
| } | |
| return kvfeed.Config{ | |
| - Writer: buf, | |
| - Settings: cfg.Settings, | |
| - DB: cfg.DB, | |
| - Codec: cfg.Codec, | |
| - Clock: cfg.DB.Clock(), | |
| - Gossip: cfg.Gossip, | |
| - Spans: spans, | |
| - BackfillCheckpoint: ca.spec.Checkpoint.Spans, | |
| - Targets: AllTargets(ca.spec.Feed), | |
| - Metrics: &ca.metrics.KVFeedMetrics, | |
| - OnBackfillCallback: ca.sliMetrics.getBackfillCallback(), | |
| - OnBackfillRangeCallback: ca.sliMetrics.getBackfillRangeCallback(), | |
| - MM: ca.kvFeedMemMon, | |
| - InitialHighWater: initialHighWater, | |
| - WithDiff: withDiff, | |
| - NeedsInitialScan: needsInitialScan, | |
| - SchemaChangeEvents: schemaChangeEvents, | |
| - SchemaChangePolicy: schemaChangePolicy, | |
| - SchemaFeed: sf, | |
| - Knobs: ca.knobs.FeedKnobs, | |
| + Writer: buf, | |
| + Settings: cfg.Settings, | |
| + DB: cfg.DB, | |
| + Codec: cfg.Codec, | |
| + Clock: cfg.DB.Clock(), | |
| + Gossip: cfg.Gossip, | |
| + Spans: spans, | |
| + BackfillCheckpoint: ca.spec.Checkpoint.Spans, | |
| + Targets: AllTargets(ca.spec.Feed), | |
| + Metrics: &ca.metrics.KVFeedMetrics, | |
| + OnBackfillCallback: ca.sliMetrics.getBackfillCallback(), | |
| + MM: ca.kvFeedMemMon, | |
| + InitialHighWater: initialHighWater, | |
| + WithDiff: withDiff, | |
| + NeedsInitialScan: needsInitialScan, | |
| + SchemaChangeEvents: schemaChangeEvents, | |
| + SchemaChangePolicy: schemaChangePolicy, | |
| + SchemaFeed: sf, | |
| + Knobs: ca.knobs.FeedKnobs, | |
| } | |
| } | |
| @@ -774,15 +768,13 @@ func (c *kvEventToRowConsumer) eventToRow( | |
| prevSchemaTimestamp = schemaTimestamp.Prev() | |
| } | |
| - desc, family, err := c.rfCache.TableDescForKey(ctx, event.KV().Key, schemaTimestamp) | |
| + desc, err := c.rfCache.TableDescForKey(ctx, event.KV().Key, schemaTimestamp) | |
| if err != nil { | |
| return r, err | |
| } | |
| r.tableDesc = desc | |
| - r.familyID = family | |
| - var rf *row.Fetcher | |
| - rf, err = c.rfCache.RowFetcherForColumnFamily(desc, family) | |
| + rf, err := c.rfCache.RowFetcherForTableDesc(desc) | |
| if err != nil { | |
| return r, err | |
| } | |
| @@ -826,20 +818,17 @@ func (c *kvEventToRowConsumer) eventToRow( | |
| if withDiff { | |
| prevRF := rf | |
| r.prevTableDesc = r.tableDesc | |
| - r.prevFamilyID = r.familyID | |
| if prevSchemaTimestamp != schemaTimestamp { | |
| // If the previous value is being interpreted under a different | |
| // version of the schema, fetch the correct table descriptor and | |
| // create a new row.Fetcher with it. | |
| - prevDesc, family, err := c.rfCache.TableDescForKey(ctx, event.KV().Key, prevSchemaTimestamp) | |
| - | |
| + prevDesc, err := c.rfCache.TableDescForKey(ctx, event.KV().Key, prevSchemaTimestamp) | |
| if err != nil { | |
| return r, err | |
| } | |
| - r.prevTableDesc = prevDesc | |
| - r.prevFamilyID = family | |
| - prevRF, err = c.rfCache.RowFetcherForColumnFamily(prevDesc, family) | |
| + r.prevTableDesc = prevDesc | |
| + prevRF, err = c.rfCache.RowFetcherForTableDesc(prevDesc) | |
| if err != nil { | |
| return r, err | |
| } | |
| @@ -1142,9 +1131,6 @@ func (cf *changeFrontier) MustBeStreaming() bool { | |
| // Start is part of the RowSource interface. | |
| func (cf *changeFrontier) Start(ctx context.Context) { | |
| - if cf.spec.JobID != 0 { | |
| - ctx = logtags.AddTag(ctx, "job", cf.spec.JobID) | |
| - } | |
| // StartInternal called at the beginning of the function because there are | |
| // early returns if errors are detected. | |
| ctx = cf.StartInternal(ctx, changeFrontierProcName) | |
| @@ -1363,7 +1349,7 @@ func (cf *changeFrontier) noteAggregatorProgress(d rowenc.EncDatum) error { | |
| if !resolved.Timestamp.IsEmpty() && resolved.Timestamp.Less(cf.highWaterAtStart) { | |
| logcrash.ReportOrPanic(cf.Ctx, &cf.flowCtx.Cfg.Settings.SV, | |
| `got a span level timestamp %s for %s that is less than the initial high-water %s`, | |
| - redact.Safe(resolved.Timestamp), resolved.Span, redact.Safe(cf.highWaterAtStart)) | |
| + log.Safe(resolved.Timestamp), resolved.Span, log.Safe(cf.highWaterAtStart)) | |
| continue | |
| } | |
| if err := cf.forwardFrontier(resolved); err != nil { | |
| diff --git a/pkg/ccl/changefeedccl/changefeed_stmt.go b/pkg/ccl/changefeedccl/changefeed_stmt.go | |
| index 78276ca28e..23012b33e9 100644 | |
| --- a/pkg/ccl/changefeedccl/changefeed_stmt.go | |
| +++ b/pkg/ccl/changefeedccl/changefeed_stmt.go | |
| @@ -112,7 +112,6 @@ func changefeedPlanHook( | |
| } | |
| } | |
| - //TODO: We're passing around the full output of optsFn a lot, make it a type. | |
| optsFn, err := p.TypeAsStringOpts(ctx, changefeedStmt.Options, changefeedbase.ChangefeedOptionExpectValues) | |
| if err != nil { | |
| return nil, nil, nil, false, err | |
| @@ -185,9 +184,11 @@ func changefeedPlanHook( | |
| jobID := p.ExecCfg().JobRegistry.MakeJobID() | |
| { | |
| var ptr *ptpb.Record | |
| + var protectedTimestampID uuid.UUID | |
| codec := p.ExecCfg().Codec | |
| if shouldProtectTimestamps(codec) { | |
| ptr = createProtectedTimestampRecord(ctx, codec, jobID, AllTargets(details), details.StatementTime, progress.GetChangefeed()) | |
| + protectedTimestampID = ptr.ID.GetUUID() | |
| } | |
| jr.Progress = *progress.GetChangefeed() | |
| @@ -208,6 +209,19 @@ func changefeedPlanHook( | |
| } | |
| return err | |
| } | |
| + // If we created a protected timestamp for an initial scan, verify it. | |
| + // Doing this synchronously here rather than asynchronously later provides | |
| + // a nice UX win in the case that the data isn't actually available. | |
| + if protectedTimestampID != uuid.Nil { | |
| + if err := p.ExecCfg().ProtectedTimestampProvider.Verify(ctx, protectedTimestampID); err != nil { | |
| + if cancelErr := sj.Cancel(ctx); cancelErr != nil { | |
| + if ctx.Err() == nil { | |
| + log.Warningf(ctx, "failed to cancel job: %v", cancelErr) | |
| + } | |
| + } | |
| + return err | |
| + } | |
| + } | |
| } | |
| // Start the job. | |
| @@ -524,16 +538,12 @@ func getTargetsAndTables( | |
| tables[table.GetID()] = jobspb.ChangefeedTargetTable{ | |
| StatementTimeName: name, | |
| } | |
| - typ := jobspb.ChangefeedTargetSpecification_PRIMARY_FAMILY_ONLY | |
| - if len(table.GetFamilies()) > 1 { | |
| - typ = jobspb.ChangefeedTargetSpecification_EACH_FAMILY | |
| - } | |
| ts := jobspb.ChangefeedTargetSpecification{ | |
| - Type: typ, | |
| + Type: jobspb.ChangefeedTargetSpecification_PRIMARY_FAMILY_ONLY, | |
| TableID: table.GetID(), | |
| } | |
| targets = append(targets, ts) | |
| - if err := changefeedbase.ValidateTable(targets, table, opts); err != nil { | |
| + if err := changefeedbase.ValidateTable(targets, table); err != nil { | |
| return nil, nil, err | |
| } | |
| for _, warning := range changefeedbase.WarningsForTable(tables, table, opts) { | |
| diff --git a/pkg/ccl/changefeedccl/changefeed_test.go b/pkg/ccl/changefeedccl/changefeed_test.go | |
| index e64b007e6a..04f1ccb6ab 100644 | |
| --- a/pkg/ccl/changefeedccl/changefeed_test.go | |
| +++ b/pkg/ccl/changefeedccl/changefeed_test.go | |
| @@ -54,6 +54,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/server" | |
| "github.com/cockroachdb/cockroach/pkg/server/telemetry" | |
| "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| + "github.com/cockroachdb/cockroach/pkg/sql" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/bootstrap" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/descpb" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/desctestutils" | |
| @@ -799,77 +800,6 @@ func TestChangefeedInitialScan(t *testing.T) { | |
| t.Run(`pubsub`, pubsubTest(testFn)) | |
| } | |
| -func TestChangefeedBackfillObservability(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| - sqlDB := sqlutils.MakeSQLRunner(db) | |
| - | |
| - knobs := f.Server().(*server.TestServer).Cfg.TestingKnobs. | |
| - DistSQL.(*execinfra.TestingKnobs). | |
| - Changefeed.(*TestingKnobs) | |
| - | |
| - registry := f.Server().JobRegistry().(*jobs.Registry) | |
| - sli, err := registry.MetricsStruct().Changefeed.(*Metrics).getSLIMetrics(defaultSLIScope) | |
| - require.NoError(t, err) | |
| - pendingRanges := sli.BackfillPendingRanges | |
| - | |
| - // Create a table with multiple ranges | |
| - numRanges := 10 | |
| - rowsPerRange := 20 | |
| - sqlDB.Exec(t, fmt.Sprintf(` | |
| - CREATE TABLE foo (key INT PRIMARY KEY); | |
| - INSERT INTO foo (key) SELECT * FROM generate_series(1, %d); | |
| - ALTER TABLE foo SPLIT AT (SELECT * FROM generate_series(%d, %d, %d)); | |
| - `, numRanges*rowsPerRange, rowsPerRange, (numRanges-1)*rowsPerRange, rowsPerRange)) | |
| - sqlDB.CheckQueryResults(t, `SELECT count(*) FROM [SHOW RANGES FROM TABLE foo]`, | |
| - [][]string{{fmt.Sprint(numRanges)}}, | |
| - ) | |
| - | |
| - // Allow control of the scans | |
| - scanCtx, scanCancel := context.WithCancel(context.Background()) | |
| - scanChan := make(chan struct{}) | |
| - knobs.FeedKnobs.BeforeScanRequest = func(b *kv.Batch) error { | |
| - select { | |
| - case <-scanCtx.Done(): | |
| - return scanCtx.Err() | |
| - case <-scanChan: | |
| - return nil | |
| - } | |
| - } | |
| - | |
| - require.Equal(t, pendingRanges.Value(), int64(0)) | |
| - foo := feed(t, f, `CREATE CHANGEFEED FOR foo`) | |
| - defer closeFeed(t, foo) | |
| - | |
| - // Progress the initial backfill halfway through its ranges | |
| - for i := 0; i < numRanges/2; i++ { | |
| - scanChan <- struct{}{} | |
| - } | |
| - testutils.SucceedsSoon(t, func() error { | |
| - count := pendingRanges.Value() | |
| - if count != int64(numRanges/2) { | |
| - return fmt.Errorf("range count %d should be %d", count, numRanges/2) | |
| - } | |
| - return nil | |
| - }) | |
| - | |
| - // Ensure that the pending count is cleared if the backfill completes | |
| - // regardless of successful scans | |
| - scanCancel() | |
| - testutils.SucceedsSoon(t, func() error { | |
| - count := pendingRanges.Value() | |
| - if count > 0 { | |
| - return fmt.Errorf("range count %d should be 0", count) | |
| - } | |
| - return nil | |
| - }) | |
| - } | |
| - | |
| - t.Run("enterprise", enterpriseTest(testFn, feedTestNoTenants)) | |
| -} | |
| - | |
| func TestChangefeedUserDefinedTypes(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| @@ -1202,9 +1132,8 @@ func TestChangefeedSchemaChangeBackfillCheckpoint(t *testing.T) { | |
| // Ensure Scan Requests are always small enough that we receive multiple | |
| // resolved events during a backfill | |
| - knobs.FeedKnobs.BeforeScanRequest = func(b *kv.Batch) error { | |
| + knobs.FeedKnobs.BeforeScanRequest = func(b *kv.Batch) { | |
| b.Header.MaxSpanRequestKeys = 10 | |
| - return nil | |
| } | |
| // Setup changefeed job details, avoid relying on initial scan functionality | |
| @@ -1773,48 +1702,15 @@ func TestChangefeedColumnFamily(t *testing.T) { | |
| defer log.Scope(t).Close(t) | |
| testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| - | |
| sqlDB := sqlutils.MakeSQLRunner(db) | |
| // Table with 2 column families. | |
| - sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING, c STRING, FAMILY (a,b), FAMILY (c))`) | |
| - sqlDB.Exec(t, `INSERT INTO foo values (0, 'dog', 'cat')`) | |
| - | |
| - // Must specify WITH split_column_families | |
| - sqlDB.ExpectErr(t, `multiple column families`, `CREATE CHANGEFEED FOR foo`) | |
| - | |
| - foo := feed(t, f, `CREATE CHANGEFEED FOR foo WITH split_column_families`) | |
| - defer closeFeed(t, foo) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"a": 0, "b": "dog"}}`, | |
| - `foo: [0]->{"after": {"c": "cat"}}`, | |
| - }) | |
| - | |
| - // No messages for unaffected column families. | |
| - sqlDB.Exec(t, `UPDATE foo SET c='lion' WHERE a=0`) | |
| - sqlDB.Exec(t, `UPDATE foo SET c='tiger' WHERE a=0`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"c": "lion"}}`, | |
| - `foo: [0]->{"after": {"c": "tiger"}}`, | |
| - }) | |
| - | |
| - // No messages on insert for families where no non-null values were set. | |
| - sqlDB.Exec(t, `INSERT INTO foo values (1, 'puppy', null)`) | |
| - sqlDB.Exec(t, `INSERT INTO foo values (2, null, 'kitten')`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [1]->{"after": {"a": 1, "b": "puppy"}}`, | |
| - `foo: [2]->{"after": {"a": 2, "b": null}}`, | |
| - `foo: [2]->{"after": {"c": "kitten"}}`, | |
| - }) | |
| - | |
| - // Deletes send a message for each column family. | |
| - fooWithDiff := feed(t, f, `CREATE CHANGEFEED FOR foo WITH split_column_families, diff, no_initial_scan, resolved='1s'`) | |
| - defer closeFeed(t, fooWithDiff) | |
| - sqlDB.Exec(t, `DELETE FROM foo WHERE a=0`) | |
| - assertPayloads(t, fooWithDiff, []string{ | |
| - `foo: [0]->{"after": null, "before": {"a": 0, "b": "dog"}}`, | |
| - `foo: [0]->{"after": null, "before": {"c": "tiger"}}`, | |
| - }) | |
| + sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING, FAMILY (a), FAMILY (b))`) | |
| + if strings.Contains(t.Name(), `enterprise`) { | |
| + sqlDB.ExpectErr(t, `exactly 1 column family`, `CREATE CHANGEFEED FOR foo`) | |
| + } else { | |
| + sqlDB.ExpectErr(t, `exactly 1 column family`, `EXPERIMENTAL CHANGEFEED FOR foo`) | |
| + } | |
| // Table with a second column family added after the changefeed starts. | |
| sqlDB.Exec(t, `CREATE TABLE bar (a INT PRIMARY KEY, FAMILY f_a (a))`) | |
| @@ -1826,116 +1722,16 @@ func TestChangefeedColumnFamily(t *testing.T) { | |
| }) | |
| sqlDB.Exec(t, `ALTER TABLE bar ADD COLUMN b STRING CREATE FAMILY f_b`) | |
| sqlDB.Exec(t, `INSERT INTO bar VALUES (1)`) | |
| - if _, err := bar.Next(); !testutils.IsError(err, `created on a table with a single column family`) { | |
| - t.Errorf(`expected "column family" error got: %+v`, err) | |
| + if _, err := bar.Next(); !testutils.IsError(err, `exactly 1 column family`) { | |
| + t.Errorf(`expected "exactly 1 column family" error got: %+v`, err) | |
| } | |
| } | |
| t.Run(`sinkless`, sinklessTest(testFn)) | |
| t.Run(`enterprise`, enterpriseTest(testFn)) | |
| t.Run(`kafka`, kafkaTest(testFn)) | |
| -} | |
| - | |
| -func TestChangefeedEachColumnFamilySchemaChanges(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| - | |
| - sqlDB := sqlutils.MakeSQLRunner(db) | |
| - | |
| - // Table with 2 column families. | |
| - sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING, c STRING, FAMILY f1 (a,b), FAMILY f2 (c))`) | |
| - sqlDB.Exec(t, `INSERT INTO foo values (0, 'dog', 'cat')`) | |
| - foo := feed(t, f, `CREATE CHANGEFEED FOR foo WITH split_column_families`) | |
| - defer closeFeed(t, foo) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"a": 0, "b": "dog"}}`, | |
| - `foo: [0]->{"after": {"c": "cat"}}`, | |
| - }) | |
| - | |
| - // Add a column to an existing family | |
| - sqlDB.Exec(t, `ALTER TABLE foo ADD COLUMN d string DEFAULT 'hi' FAMILY f2`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"c": "cat", "d": "hi"}}`, | |
| - }) | |
| - | |
| - // Add a column to a new family. | |
| - // Behavior here is a little wonky with default values in a way | |
| - // that's likely to change with declarative schema changer, | |
| - // so not asserting anything either way about that. | |
| - sqlDB.Exec(t, `ALTER TABLE foo ADD COLUMN e string CREATE FAMILY f3`) | |
| - sqlDB.Exec(t, `UPDATE foo SET e='hello' WHERE a=0`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"e": "hello"}}`, | |
| - }) | |
| - | |
| - } | |
| - | |
| - t.Run(`sinkless`, sinklessTest(testFn)) | |
| - t.Run(`enterprise`, enterpriseTest(testFn)) | |
| - t.Run(`kafka`, kafkaTest(testFn)) | |
| -} | |
| - | |
| -func TestChangefeedEachColumnFamilyVirtualColumns(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| - | |
| - sqlDB := sqlutils.MakeSQLRunner(db) | |
| - | |
| - // Table with 2 column families. | |
| - sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING, c STRING, FAMILY f1 (a,b), FAMILY f2 (c))`) | |
| - sqlDB.Exec(t, `INSERT INTO foo values (0, 'dog', 'cat')`) | |
| - foo := feed(t, f, `CREATE CHANGEFEED FOR foo WITH split_column_families`) | |
| - defer closeFeed(t, foo) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"a": 0, "b": "dog"}}`, | |
| - `foo: [0]->{"after": {"c": "cat"}}`, | |
| - }) | |
| - | |
| - // Add a column to an existing family | |
| - sqlDB.Exec(t, `ALTER TABLE foo ADD COLUMN d string DEFAULT 'hi' FAMILY f2`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"c": "cat", "d": "hi"}}`, | |
| - }) | |
| - | |
| - // Add a column to a new family. | |
| - // Behavior here is a little wonky with default values in a way | |
| - // that's likely to change with declarative schema changer, | |
| - // so not asserting anything either way about that. | |
| - sqlDB.Exec(t, `ALTER TABLE foo ADD COLUMN e string CREATE FAMILY f3`) | |
| - sqlDB.Exec(t, `UPDATE foo SET e='hello' WHERE a=0`) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: [0]->{"after": {"e": "hello"}}`, | |
| - }) | |
| - | |
| - } | |
| - | |
| - t.Run(`sinkless`, sinklessTest(testFn)) | |
| - t.Run(`enterprise`, enterpriseTest(testFn)) | |
| - t.Run(`kafka`, kafkaTest(testFn)) | |
| -} | |
| - | |
| -func TestChangefeedColumnFamilyAvro(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| - sqlDB := sqlutils.MakeSQLRunner(db) | |
| - | |
| - sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING, c STRING, FAMILY most (a,b), FAMILY justc (c))`) | |
| - sqlDB.Exec(t, `INSERT INTO foo values (0, 'dog', 'cat')`) | |
| - foo := feed(t, f, `CREATE CHANGEFEED FOR foo WITH split_column_families, format=avro`) | |
| - defer closeFeed(t, foo) | |
| - assertPayloads(t, foo, []string{ | |
| - `foo: {"a":{"long":0}}->{"after":{"foomost":{"a":{"long":0},"b":{"string":"dog"}}}}`, | |
| - `foo: {"a":{"long":0}}->{"after":{"foojustc":{"c":{"string":"cat"}}}}`, | |
| - }) | |
| - | |
| - } | |
| - t.Run(`kafka`, kafkaTest(testFn)) | |
| + t.Run(`webhook`, webhookTest(testFn)) | |
| + t.Run(`pubsub`, pubsubTest(testFn)) | |
| } | |
| func TestChangefeedAuthorization(t *testing.T) { | |
| @@ -3128,13 +2924,12 @@ func TestChangefeedDataTTL(t *testing.T) { | |
| knobs := f.Server().TestingKnobs(). | |
| DistSQL.(*execinfra.TestingKnobs). | |
| Changefeed.(*TestingKnobs) | |
| - knobs.FeedKnobs.BeforeScanRequest = func(_ *kv.Batch) error { | |
| + knobs.FeedKnobs.BeforeScanRequest = func(_ *kv.Batch) { | |
| if atomic.LoadInt32(&shouldWait) == 0 { | |
| - return nil | |
| + return | |
| } | |
| wait <- struct{}{} | |
| <-resume | |
| - return nil | |
| } | |
| sqlDB := sqlutils.MakeSQLRunner(db) | |
| @@ -4265,6 +4060,67 @@ func TestChangefeedProtectedTimestampOnPause(t *testing.T) { | |
| } | |
| +// This test ensures that the changefeed attempts to verify its initial protected | |
| +// timestamp record and that when that verification fails, the job is canceled | |
| +// and the record removed. | |
| +func TestChangefeedProtectedTimestampsVerificationFails(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + defer log.Scope(t).Close(t) | |
| + | |
| + verifyRequestCh := make(chan *roachpb.AdminVerifyProtectedTimestampRequest, 1) | |
| + requestFilter := kvserverbase.ReplicaRequestFilter(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) *roachpb.Error { | |
| + if r, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + req := r.(*roachpb.AdminVerifyProtectedTimestampRequest) | |
| + verifyRequestCh <- req | |
| + return roachpb.NewError(errors.Errorf("failed to verify protection %v on %v", req.RecordID, ba.RangeID)) | |
| + } | |
| + return nil | |
| + }) | |
| + | |
| + setStoreKnobs := func(args *base.TestServerArgs) { | |
| + storeKnobs := &kvserver.StoreTestingKnobs{} | |
| + storeKnobs.TestingRequestFilter = requestFilter | |
| + args.Knobs.Store = storeKnobs | |
| + } | |
| + | |
| + testFn := func(t *testing.T, db *gosql.DB, f cdctest.TestFeedFactory) { | |
| + ctx := context.Background() | |
| + sqlDB := sqlutils.MakeSQLRunner(db) | |
| + sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING)`) | |
| + _, err := f.Feed(`CREATE CHANGEFEED FOR foo WITH resolved`) | |
| + // Make sure we got the injected error. | |
| + require.Regexp(t, "failed to verify", err) | |
| + // Make sure we tried to verify the request. | |
| + r := <-verifyRequestCh | |
| + cfg := f.Server().ExecutorConfig().(sql.ExecutorConfig) | |
| + kvDB := cfg.DB | |
| + pts := cfg.ProtectedTimestampProvider | |
| + // Make sure that the canceled job gets moved through its OnFailOrCancel | |
| + // phase and removes its protected timestamp. | |
| + testutils.SucceedsSoon(t, func() error { | |
| + err := kvDB.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| + _, err := pts.GetRecord(ctx, txn, r.RecordID) | |
| + return err | |
| + }) | |
| + if err == nil { | |
| + return errors.Errorf("expected record to be removed") | |
| + } | |
| + if errors.Is(err, protectedts.ErrNotExists) { | |
| + return nil | |
| + } | |
| + return err | |
| + }) | |
| + } | |
| + opts := []feedTestOption{withArgsFn(setStoreKnobs), feedTestNoTenants} | |
| + t.Run(`enterprise`, enterpriseTest(testFn, opts...)) | |
| + t.Run(`cloudstorage`, cloudStorageTest(testFn, opts...)) | |
| + t.Run(`kafka`, kafkaTest(testFn, opts...)) | |
| + t.Run(`webhook`, webhookTest(testFn, opts...)) | |
| + t.Run(`pubsub`, pubsubTest(testFn, opts...)) | |
| +} | |
| + | |
| func TestManyChangefeedsOneTable(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| @@ -4371,8 +4227,8 @@ func TestChangefeedNodeShutdown(t *testing.T) { | |
| defer tc.Stopper().Stop(context.Background()) | |
| db := tc.ServerConn(1) | |
| - serverutils.SetClusterSetting(t, tc, "changefeed.experimental_poll_interval", time.Millisecond) | |
| sqlDB := sqlutils.MakeSQLRunner(db) | |
| + sqlDB.Exec(t, `SET CLUSTER SETTING changefeed.experimental_poll_interval = '0ns'`) | |
| sqlDB.Exec(t, `CREATE DATABASE d`) | |
| sqlDB.Exec(t, `CREATE TABLE foo (a INT PRIMARY KEY, b STRING)`) | |
| sqlDB.Exec(t, `INSERT INTO foo VALUES (0, 'initial')`) | |
| @@ -4583,22 +4439,19 @@ func TestChangefeedHandlesDrainingNodes(t *testing.T) { | |
| skip.UnderRace(t, "Takes too long with race enabled") | |
| shouldDrain := true | |
| - knobs := base.TestingKnobs{ | |
| - DistSQL: &execinfra.TestingKnobs{ | |
| - DrainFast: true, | |
| - Changefeed: &TestingKnobs{}, | |
| - Flowinfra: &flowinfra.TestingKnobs{ | |
| - FlowRegistryDraining: func() bool { | |
| - if shouldDrain { | |
| - shouldDrain = false | |
| - return true | |
| - } | |
| - return false | |
| - }, | |
| + knobs := base.TestingKnobs{DistSQL: &execinfra.TestingKnobs{ | |
| + DrainFast: true, | |
| + Changefeed: &TestingKnobs{}, | |
| + Flowinfra: &flowinfra.TestingKnobs{ | |
| + FlowRegistryDraining: func() bool { | |
| + if shouldDrain { | |
| + shouldDrain = false | |
| + return true | |
| + } | |
| + return false | |
| }, | |
| }, | |
| - JobsTestingKnobs: jobs.NewTestingKnobsWithShortIntervals(), | |
| - } | |
| + }} | |
| sinkDir, cleanupFn := testutils.TempDir(t) | |
| defer cleanupFn() | |
| @@ -4613,9 +4466,9 @@ func TestChangefeedHandlesDrainingNodes(t *testing.T) { | |
| db := tc.ServerConn(1) | |
| sqlDB := sqlutils.MakeSQLRunner(db) | |
| - serverutils.SetClusterSetting(t, tc, "kv.rangefeed.enabled", true) | |
| - serverutils.SetClusterSetting(t, tc, "kv.closed_timestamp.target_duration", time.Second) | |
| - serverutils.SetClusterSetting(t, tc, "changefeed.experimental_poll_interval", 10*time.Millisecond) | |
| + sqlDB.Exec(t, `SET CLUSTER SETTING kv.rangefeed.enabled = true`) | |
| + sqlDB.Exec(t, `SET CLUSTER SETTING kv.closed_timestamp.target_duration = '1s'`) | |
| + sqlDB.Exec(t, `SET CLUSTER SETTING changefeed.experimental_poll_interval = '10ms'`) | |
| sqlutils.CreateTable( | |
| t, db, "foo", | |
| @@ -4638,9 +4491,9 @@ func TestChangefeedHandlesDrainingNodes(t *testing.T) { | |
| defer closeFeed(t, feed) | |
| // At this point, the job created by feed will fail to start running on node 0 due to draining | |
| - // registry. However, this job will be retried, and it should succeed. | |
| + // registry. However, this job will be retried, and it should succeeded. | |
| // Note: This test is a bit unrealistic in that if the registry is draining, that | |
| - // means that the server is draining (i.e. being shut down). We don't do a full shutdown | |
| + // means that the server is draining (i.e being shut down). We don't do a full shutdown | |
| // here, but we are simulating a restart by failing to start a flow the first time around. | |
| assertPayloads(t, feed, []string{ | |
| `foo: [1]->{"after": {"k": 1, "v": 1}}`, | |
| @@ -5034,9 +4887,8 @@ func TestChangefeedBackfillCheckpoint(t *testing.T) { | |
| // Ensure Scan Requests are always small enough that we receive multiple | |
| // resolved events during a backfill | |
| - knobs.FeedKnobs.BeforeScanRequest = func(b *kv.Batch) error { | |
| + knobs.FeedKnobs.BeforeScanRequest = func(b *kv.Batch) { | |
| b.Header.MaxSpanRequestKeys = 1 + rnd.Int63n(100) | |
| - return nil | |
| } | |
| // Emit resolved events for majority of spans. Be extra paranoid and ensure that | |
| diff --git a/pkg/ccl/changefeedccl/changefeedbase/options.go b/pkg/ccl/changefeedccl/changefeedbase/options.go | |
| index f161163176..83b5c54d1e 100644 | |
| --- a/pkg/ccl/changefeedccl/changefeedbase/options.go | |
| +++ b/pkg/ccl/changefeedccl/changefeedbase/options.go | |
| @@ -49,7 +49,6 @@ const ( | |
| OptCompression = `compression` | |
| OptSchemaChangeEvents = `schema_change_events` | |
| OptSchemaChangePolicy = `schema_change_policy` | |
| - OptSplitColumnFamilies = `split_column_families` | |
| OptProtectDataFromGCOnPause = `protect_data_from_gc_on_pause` | |
| OptWebhookAuthHeader = `webhook_auth_header` | |
| OptWebhookClientTimeout = `webhook_client_timeout` | |
| @@ -179,7 +178,6 @@ var ChangefeedOptionExpectValues = map[string]sql.KVStringOptValidate{ | |
| OptCompression: sql.KVStringOptRequireValue, | |
| OptSchemaChangeEvents: sql.KVStringOptRequireValue, | |
| OptSchemaChangePolicy: sql.KVStringOptRequireValue, | |
| - OptSplitColumnFamilies: sql.KVStringOptRequireNoValue, | |
| OptInitialScan: sql.KVStringOptRequireNoValue, | |
| OptNoInitialScan: sql.KVStringOptRequireNoValue, | |
| OptProtectDataFromGCOnPause: sql.KVStringOptRequireNoValue, | |
| @@ -205,7 +203,7 @@ var CommonOptions = makeStringSet(OptCursor, OptEnvelope, | |
| OptFormat, OptFullTableName, | |
| OptKeyInValue, OptTopicInValue, | |
| OptResolvedTimestamps, OptUpdatedTimestamps, | |
| - OptMVCCTimestamps, OptDiff, OptSplitColumnFamilies, | |
| + OptMVCCTimestamps, OptDiff, | |
| OptSchemaChangeEvents, OptSchemaChangePolicy, | |
| OptProtectDataFromGCOnPause, OptOnError, | |
| OptInitialScan, OptNoInitialScan, | |
| diff --git a/pkg/ccl/changefeedccl/changefeedbase/validate.go b/pkg/ccl/changefeedccl/changefeedbase/validate.go | |
| index fb5facb737..056cab49ed 100644 | |
| --- a/pkg/ccl/changefeedccl/changefeedbase/validate.go | |
| +++ b/pkg/ccl/changefeedccl/changefeedbase/validate.go | |
| @@ -16,9 +16,7 @@ import ( | |
| // ValidateTable validates that a table descriptor can be watched by a CHANGEFEED. | |
| func ValidateTable( | |
| - targets []jobspb.ChangefeedTargetSpecification, | |
| - tableDesc catalog.TableDescriptor, | |
| - opts map[string]string, | |
| + targets []jobspb.ChangefeedTargetSpecification, tableDesc catalog.TableDescriptor, | |
| ) error { | |
| var t jobspb.ChangefeedTargetSpecification | |
| var found bool | |
| @@ -50,17 +48,11 @@ func ValidateTable( | |
| if tableDesc.IsSequence() { | |
| return errors.Errorf(`CHANGEFEED cannot target sequences: %s`, tableDesc.GetName()) | |
| } | |
| - if t.Type == jobspb.ChangefeedTargetSpecification_PRIMARY_FAMILY_ONLY && len(tableDesc.GetFamilies()) != 1 { | |
| + if len(tableDesc.GetFamilies()) != 1 { | |
| return errors.Errorf( | |
| - `CHANGEFEED created on a table with a single column family (%s) cannot now target a table with %d families.`, | |
| + `CHANGEFEEDs are currently supported on tables with exactly 1 column family: %s has %d`, | |
| tableDesc.GetName(), len(tableDesc.GetFamilies())) | |
| } | |
| - _, columnFamiliesOpt := opts[OptSplitColumnFamilies] | |
| - if !columnFamiliesOpt && len(tableDesc.GetFamilies()) != 1 { | |
| - return errors.Errorf( | |
| - `CHANGEFEED targeting a table (%s) with multiple column families requires WITH %s and will emit multiple events per row.`, | |
| - tableDesc.GetName(), OptSplitColumnFamilies) | |
| - } | |
| if tableDesc.Dropped() { | |
| return errors.Errorf(`"%s" was dropped`, t.StatementTimeName) | |
| @@ -87,11 +79,5 @@ func WarningsForTable( | |
| } | |
| } | |
| } | |
| - if tableDesc.NumFamilies() > 1 { | |
| - warnings = append(warnings, | |
| - errors.Errorf("Table %s has %d underlying column families. Messages will be emitted separately for each family.", | |
| - tableDesc.GetName(), tableDesc.NumFamilies(), | |
| - )) | |
| - } | |
| return warnings | |
| } | |
| diff --git a/pkg/ccl/changefeedccl/encoder.go b/pkg/ccl/changefeedccl/encoder.go | |
| index 85df7113f2..7aef94121f 100644 | |
| --- a/pkg/ccl/changefeedccl/encoder.go | |
| +++ b/pkg/ccl/changefeedccl/encoder.go | |
| @@ -53,9 +53,6 @@ type encodeRow struct { | |
| // tableDesc is a TableDescriptor for the table containing `datums`. | |
| // It's valid for interpreting the row at `updated`. | |
| tableDesc catalog.TableDescriptor | |
| - // familyID indicates which column family is populated on this row. | |
| - // It's valid for interpreting the row at `updated`. | |
| - familyID descpb.FamilyID | |
| // prevDatums is the old value of a changed table row. The field is set | |
| // to nil if the before value for changes was not requested (OptDiff). | |
| prevDatums rowenc.EncDatumRow | |
| @@ -64,8 +61,6 @@ type encodeRow struct { | |
| // prevTableDesc is a TableDescriptor for the table containing `prevDatums`. | |
| // It's valid for interpreting the row at `updated.Prev()`. | |
| prevTableDesc catalog.TableDescriptor | |
| - // prevFamilyID indicates which column family is populated in prevDatums. | |
| - prevFamilyID descpb.FamilyID | |
| } | |
| // Encoder turns a row into a serialized changefeed key, value, or resolved | |
| @@ -191,23 +186,9 @@ func (e *jsonEncoder) encodeKeyRaw(row encodeRow) ([]interface{}, error) { | |
| func (e *jsonEncoder) encodeTopicRaw(row encodeRow) (interface{}, error) { | |
| descID := row.tableDesc.GetID() | |
| // use the target list since row.tableDesc.GetName() will not have fully qualified names | |
| - for _, target := range e.targets { | |
| - if target.TableID == descID { | |
| - switch target.Type { | |
| - case jobspb.ChangefeedTargetSpecification_PRIMARY_FAMILY_ONLY: | |
| - return target.StatementTimeName, nil | |
| - case jobspb.ChangefeedTargetSpecification_EACH_FAMILY: | |
| - family, err := row.tableDesc.FindFamilyByID(row.familyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - return fmt.Sprintf("%s.%s", target.StatementTimeName, family.Name), nil | |
| - case jobspb.ChangefeedTargetSpecification_COLUMN_FAMILY: | |
| - // Not implemented yet | |
| - default: | |
| - // fall through to error | |
| - } | |
| - return nil, errors.AssertionFailedf("Found a matching target with unimplemented type %s", target.Type) | |
| + for _, topic := range e.targets { | |
| + if topic.TableID == descID { | |
| + return topic.StatementTimeName, nil | |
| } | |
| } | |
| return nil, fmt.Errorf("table with name %s and descriptor ID %d not found in changefeed target list", | |
| @@ -222,64 +203,48 @@ func (e *jsonEncoder) EncodeValue(_ context.Context, row encodeRow) ([]byte, err | |
| var after map[string]interface{} | |
| if !row.deleted { | |
| - family, err := row.tableDesc.FindFamilyByID(row.familyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - include := make(map[descpb.ColumnID]struct{}, len(family.ColumnIDs)) | |
| - var yes struct{} | |
| - for _, colID := range family.ColumnIDs { | |
| - include[colID] = yes | |
| - } | |
| + columns := row.tableDesc.PublicColumns() | |
| after = make(map[string]interface{}) | |
| - for i, col := range row.tableDesc.PublicColumns() { | |
| - _, inFamily := include[col.GetID()] | |
| - virtual := col.IsVirtual() && e.virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsNull) | |
| - if inFamily || virtual { | |
| - datum := row.datums[i] | |
| - if err := datum.EnsureDecoded(col.GetType(), &e.alloc); err != nil { | |
| - return nil, err | |
| - } | |
| - after[col.GetName()], err = tree.AsJSON( | |
| - datum.Datum, | |
| - sessiondatapb.DataConversionConfig{}, | |
| - time.UTC, | |
| - ) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| + for i, col := range columns { | |
| + if col.IsVirtual() && e.virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsOmitted) { | |
| + continue | |
| + } | |
| + datum := row.datums[i] | |
| + if err := datum.EnsureDecoded(col.GetType(), &e.alloc); err != nil { | |
| + return nil, err | |
| + } | |
| + var err error | |
| + after[col.GetName()], err = tree.AsJSON( | |
| + datum.Datum, | |
| + sessiondatapb.DataConversionConfig{}, | |
| + time.UTC, | |
| + ) | |
| + if err != nil { | |
| + return nil, err | |
| } | |
| } | |
| } | |
| var before map[string]interface{} | |
| if row.prevDatums != nil && !row.prevDeleted { | |
| - family, err := row.prevTableDesc.FindFamilyByID(row.prevFamilyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - include := make(map[descpb.ColumnID]struct{}) | |
| - var yes struct{} | |
| - for _, colID := range family.ColumnIDs { | |
| - include[colID] = yes | |
| - } | |
| + columns := row.prevTableDesc.PublicColumns() | |
| before = make(map[string]interface{}) | |
| - for i, col := range row.prevTableDesc.PublicColumns() { | |
| - _, inFamily := include[col.GetID()] | |
| - virtual := col.IsVirtual() && e.virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsNull) | |
| - if inFamily || virtual { | |
| - datum := row.prevDatums[i] | |
| - if err := datum.EnsureDecoded(col.GetType(), &e.alloc); err != nil { | |
| - return nil, err | |
| - } | |
| - before[col.GetName()], err = tree.AsJSON( | |
| - datum.Datum, | |
| - sessiondatapb.DataConversionConfig{}, | |
| - time.UTC, | |
| - ) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| + for i, col := range columns { | |
| + if col.IsVirtual() && e.virtualColumnVisibility == string(changefeedbase.OptVirtualColumnsOmitted) { | |
| + continue | |
| + } | |
| + datum := row.prevDatums[i] | |
| + if err := datum.EnsureDecoded(col.GetType(), &e.alloc); err != nil { | |
| + return nil, err | |
| + } | |
| + var err error | |
| + before[col.GetName()], err = tree.AsJSON( | |
| + datum.Datum, | |
| + sessiondatapb.DataConversionConfig{}, | |
| + time.UTC, | |
| + ) | |
| + if err != nil { | |
| + return nil, err | |
| } | |
| } | |
| } | |
| @@ -377,13 +342,13 @@ type confluentAvroEncoder struct { | |
| resolvedCache map[string]confluentRegisteredEnvelopeSchema | |
| } | |
| -type tableIDAndVersion struct { | |
| - tableID descpb.ID | |
| - version descpb.DescriptorVersion | |
| - familyID descpb.FamilyID | |
| -} | |
| +type tableIDAndVersion uint64 | |
| type tableIDAndVersionPair [2]tableIDAndVersion // [before, after] | |
| +func makeTableIDAndVersion(id descpb.ID, version descpb.DescriptorVersion) tableIDAndVersion { | |
| + return tableIDAndVersion(id)<<32 + tableIDAndVersion(version) | |
| +} | |
| + | |
| type confluentRegisteredKeySchema struct { | |
| schema *avroDataRecord | |
| registryID int32 | |
| @@ -459,35 +424,19 @@ func newConfluentAvroEncoder( | |
| // Get the raw SQL-formatted string for a table name | |
| // and apply full_table_name and avro_schema_prefix options | |
| -func (e *confluentAvroEncoder) rawTableName( | |
| - desc catalog.TableDescriptor, familyID descpb.FamilyID, | |
| -) (string, error) { | |
| - for _, target := range e.targets { | |
| - if target.TableID == desc.GetID() { | |
| - switch target.Type { | |
| - case jobspb.ChangefeedTargetSpecification_PRIMARY_FAMILY_ONLY: | |
| - return e.schemaPrefix + target.StatementTimeName, nil | |
| - case jobspb.ChangefeedTargetSpecification_EACH_FAMILY: | |
| - family, err := desc.FindFamilyByID(familyID) | |
| - if err != nil { | |
| - return "", err | |
| - } | |
| - return fmt.Sprintf("%s%s.%s", e.schemaPrefix, target.StatementTimeName, family.Name), nil | |
| - case jobspb.ChangefeedTargetSpecification_COLUMN_FAMILY: | |
| - // Not implemented yet | |
| - default: | |
| - // fall through to error | |
| - } | |
| - return "", errors.AssertionFailedf("Found a matching target with unimplemented type %s", target.Type) | |
| +func (e *confluentAvroEncoder) rawTableName(desc catalog.TableDescriptor) string { | |
| + for _, spec := range e.targets { | |
| + if spec.TableID == desc.GetID() { | |
| + return e.schemaPrefix + spec.StatementTimeName | |
| } | |
| } | |
| - return desc.GetName(), errors.Newf("Could not find TargetSpecification for descriptor %v", desc) | |
| + //TODO (zinger): error here | |
| + return desc.GetName() | |
| } | |
| // EncodeKey implements the Encoder interface. | |
| func (e *confluentAvroEncoder) EncodeKey(ctx context.Context, row encodeRow) ([]byte, error) { | |
| - // No familyID in the cache key for keys because it's the same schema for all families | |
| - cacheKey := tableIDAndVersion{tableID: row.tableDesc.GetID(), version: row.tableDesc.GetVersion()} | |
| + cacheKey := makeTableIDAndVersion(row.tableDesc.GetID(), row.tableDesc.GetVersion()) | |
| var registered confluentRegisteredKeySchema | |
| v, ok := e.keyCache.Get(cacheKey) | |
| @@ -496,10 +445,7 @@ func (e *confluentAvroEncoder) EncodeKey(ctx context.Context, row encodeRow) ([] | |
| registered.schema.refreshTypeMetadata(row.tableDesc) | |
| } else { | |
| var err error | |
| - tableName, err := e.rawTableName(row.tableDesc, row.familyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| + tableName := e.rawTableName(row.tableDesc) | |
| registered.schema, err = indexToAvroSchema(row.tableDesc, row.tableDesc.GetPrimaryIndex(), tableName, e.schemaPrefix) | |
| if err != nil { | |
| return nil, err | |
| @@ -532,13 +478,9 @@ func (e *confluentAvroEncoder) EncodeValue(ctx context.Context, row encodeRow) ( | |
| var cacheKey tableIDAndVersionPair | |
| if e.beforeField && row.prevTableDesc != nil { | |
| - cacheKey[0] = tableIDAndVersion{ | |
| - tableID: row.prevTableDesc.GetID(), version: row.prevTableDesc.GetVersion(), familyID: row.prevFamilyID, | |
| - } | |
| - } | |
| - cacheKey[1] = tableIDAndVersion{ | |
| - tableID: row.tableDesc.GetID(), version: row.tableDesc.GetVersion(), familyID: row.familyID, | |
| + cacheKey[0] = makeTableIDAndVersion(row.prevTableDesc.GetID(), row.prevTableDesc.GetVersion()) | |
| } | |
| + cacheKey[1] = makeTableIDAndVersion(row.tableDesc.GetID(), row.tableDesc.GetVersion()) | |
| var registered confluentRegisteredEnvelopeSchema | |
| v, ok := e.valueCache.Get(cacheKey) | |
| @@ -552,23 +494,19 @@ func (e *confluentAvroEncoder) EncodeValue(ctx context.Context, row encodeRow) ( | |
| var beforeDataSchema *avroDataRecord | |
| if e.beforeField && row.prevTableDesc != nil { | |
| var err error | |
| - beforeDataSchema, err = tableToAvroSchema(row.prevTableDesc, row.prevFamilyID, `before`, e.schemaPrefix, e.virtualColumnVisibility) | |
| + beforeDataSchema, err = tableToAvroSchema(row.prevTableDesc, `before`, e.schemaPrefix, e.virtualColumnVisibility) | |
| if err != nil { | |
| return nil, err | |
| } | |
| } | |
| - afterDataSchema, err := tableToAvroSchema(row.tableDesc, row.familyID, avroSchemaNoSuffix, e.schemaPrefix, e.virtualColumnVisibility) | |
| + afterDataSchema, err := tableToAvroSchema(row.tableDesc, avroSchemaNoSuffix, e.schemaPrefix, e.virtualColumnVisibility) | |
| if err != nil { | |
| return nil, err | |
| } | |
| opts := avroEnvelopeOpts{afterField: true, beforeField: e.beforeField, updatedField: e.updatedField} | |
| - name, err := e.rawTableName(row.tableDesc, row.familyID) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - registered.schema, err = envelopeToAvroSchema(name, opts, beforeDataSchema, afterDataSchema, e.schemaPrefix) | |
| + registered.schema, err = envelopeToAvroSchema(e.rawTableName(row.tableDesc), opts, beforeDataSchema, afterDataSchema, e.schemaPrefix) | |
| if err != nil { | |
| return nil, err | |
| @@ -576,7 +514,7 @@ func (e *confluentAvroEncoder) EncodeValue(ctx context.Context, row encodeRow) ( | |
| // NB: This uses the kafka name escaper because it has to match the name | |
| // of the kafka topic. | |
| - subject := SQLNameToKafkaName(name) + confluentSubjectSuffixValue | |
| + subject := SQLNameToKafkaName(e.rawTableName(row.tableDesc)) + confluentSubjectSuffixValue | |
| registered.registryID, err = e.register(ctx, ®istered.schema.avroRecord, subject) | |
| if err != nil { | |
| return nil, err | |
| diff --git a/pkg/ccl/changefeedccl/encoder_test.go b/pkg/ccl/changefeedccl/encoder_test.go | |
| index c6821e0403..69b0fb8407 100644 | |
| --- a/pkg/ccl/changefeedccl/encoder_test.go | |
| +++ b/pkg/ccl/changefeedccl/encoder_test.go | |
| @@ -672,26 +672,6 @@ func TestAvroSchemaNaming(t *testing.T) { | |
| //Both changes to the subject are also reflected in the schema name in the posted schemas | |
| require.Contains(t, foo.registry.SchemaForSubject(`supermovr.public.drivers-key`), `supermovr`) | |
| require.Contains(t, foo.registry.SchemaForSubject(`supermovr.public.drivers-value`), `supermovr`) | |
| - | |
| - sqlDB.Exec(t, `ALTER TABLE movr.drivers ADD COLUMN vehicle_id int CREATE FAMILY volatile`) | |
| - multiFamilyFeed := feed(t, f, fmt.Sprintf(`CREATE CHANGEFEED FOR movr.drivers `+ | |
| - `WITH format=%s, %s`, changefeedbase.OptFormatAvro, changefeedbase.OptSplitColumnFamilies)) | |
| - defer closeFeed(t, multiFamilyFeed) | |
| - foo = multiFamilyFeed.(*kafkaFeed) | |
| - | |
| - sqlDB.Exec(t, `UPDATE movr.drivers SET vehicle_id = 1 WHERE id=1`) | |
| - | |
| - assertPayloads(t, multiFamilyFeed, []string{ | |
| - `drivers: {"id":{"long":1}}->{"after":{"driversprimary":{"id":{"long":1},"name":{"string":"Alice"}}}}`, | |
| - `drivers: {"id":{"long":1}}->{"after":{"driversvolatile":{"vehicle_id":{"long":1}}}}`, | |
| - }) | |
| - | |
| - assertRegisteredSubjects(t, foo.registry, []string{ | |
| - `drivers.primary-key`, | |
| - `drivers.primary-value`, | |
| - `drivers.volatile-value`, | |
| - }) | |
| - | |
| } | |
| t.Run(`kafka`, kafkaTest(testFn)) | |
| diff --git a/pkg/ccl/changefeedccl/kvfeed/kv_feed.go b/pkg/ccl/changefeedccl/kvfeed/kv_feed.go | |
| index ad347d7337..55751564d9 100644 | |
| --- a/pkg/ccl/changefeedccl/kvfeed/kv_feed.go | |
| +++ b/pkg/ccl/changefeedccl/kvfeed/kv_feed.go | |
| @@ -36,23 +36,22 @@ import ( | |
| // Config configures a kvfeed. | |
| type Config struct { | |
| - Settings *cluster.Settings | |
| - DB *kv.DB | |
| - Codec keys.SQLCodec | |
| - Clock *hlc.Clock | |
| - Gossip gossip.OptionalGossip | |
| - Spans []roachpb.Span | |
| - BackfillCheckpoint []roachpb.Span | |
| - Targets []jobspb.ChangefeedTargetSpecification | |
| - Writer kvevent.Writer | |
| - Metrics *kvevent.Metrics | |
| - OnBackfillCallback func() func() | |
| - OnBackfillRangeCallback func(int64) (func(), func()) | |
| - MM *mon.BytesMonitor | |
| - WithDiff bool | |
| - SchemaChangeEvents changefeedbase.SchemaChangeEventClass | |
| - SchemaChangePolicy changefeedbase.SchemaChangePolicy | |
| - SchemaFeed schemafeed.SchemaFeed | |
| + Settings *cluster.Settings | |
| + DB *kv.DB | |
| + Codec keys.SQLCodec | |
| + Clock *hlc.Clock | |
| + Gossip gossip.OptionalGossip | |
| + Spans []roachpb.Span | |
| + BackfillCheckpoint []roachpb.Span | |
| + Targets []jobspb.ChangefeedTargetSpecification | |
| + Writer kvevent.Writer | |
| + Metrics *kvevent.Metrics | |
| + OnBackfillCallback func() func() | |
| + MM *mon.BytesMonitor | |
| + WithDiff bool | |
| + SchemaChangeEvents changefeedbase.SchemaChangeEventClass | |
| + SchemaChangePolicy changefeedbase.SchemaChangePolicy | |
| + SchemaFeed schemafeed.SchemaFeed | |
| // If true, the feed will begin with a dump of data at exactly the | |
| // InitialHighWater. This is a peculiar behavior. In general the | |
| @@ -75,10 +74,9 @@ func Run(ctx context.Context, cfg Config) error { | |
| var sc kvScanner | |
| { | |
| sc = &scanRequestScanner{ | |
| - settings: cfg.Settings, | |
| - gossip: cfg.Gossip, | |
| - db: cfg.DB, | |
| - onBackfillRangeCallback: cfg.OnBackfillRangeCallback, | |
| + settings: cfg.Settings, | |
| + gossip: cfg.Gossip, | |
| + db: cfg.DB, | |
| } | |
| } | |
| var pff physicalFeedFactory | |
| diff --git a/pkg/ccl/changefeedccl/kvfeed/scanner.go b/pkg/ccl/changefeedccl/kvfeed/scanner.go | |
| index c33398144c..5bfbeb67de 100644 | |
| --- a/pkg/ccl/changefeedccl/kvfeed/scanner.go | |
| +++ b/pkg/ccl/changefeedccl/kvfeed/scanner.go | |
| @@ -40,10 +40,9 @@ type kvScanner interface { | |
| } | |
| type scanRequestScanner struct { | |
| - settings *cluster.Settings | |
| - gossip gossip.OptionalGossip | |
| - db *kv.DB | |
| - onBackfillRangeCallback func(int64) (func(), func()) | |
| + settings *cluster.Settings | |
| + gossip gossip.OptionalGossip | |
| + db *kv.DB | |
| } | |
| var _ kvScanner = (*scanRequestScanner)(nil) | |
| @@ -66,12 +65,6 @@ func (p *scanRequestScanner) Scan( | |
| return err | |
| } | |
| - var backfillDec, backfillClear func() | |
| - if p.onBackfillRangeCallback != nil { | |
| - backfillDec, backfillClear = p.onBackfillRangeCallback(int64(len(spans))) | |
| - defer backfillClear() | |
| - } | |
| - | |
| maxConcurrentScans := maxConcurrentScanRequests(p.gossip, &p.settings.SV) | |
| exportLim := limit.MakeConcurrentRequestLimiter("changefeedScanRequestLimiter", maxConcurrentScans) | |
| @@ -99,9 +92,6 @@ func (p *scanRequestScanner) Scan( | |
| defer limAlloc.Release() | |
| err := p.exportSpan(ctx, span, cfg.Timestamp, cfg.WithDiff, sink, cfg.Knobs) | |
| finished := atomic.AddInt64(&atomicFinished, 1) | |
| - if backfillDec != nil { | |
| - backfillDec() | |
| - } | |
| if log.V(2) { | |
| log.Infof(ctx, `exported %d of %d: %v`, finished, len(spans), err) | |
| } | |
| @@ -140,9 +130,7 @@ func (p *scanRequestScanner) exportSpan( | |
| // during result parsing. | |
| b.AddRawRequest(r) | |
| if knobs.BeforeScanRequest != nil { | |
| - if err := knobs.BeforeScanRequest(b); err != nil { | |
| - return err | |
| - } | |
| + knobs.BeforeScanRequest(b) | |
| } | |
| if err := txn.Run(ctx, b); err != nil { | |
| diff --git a/pkg/ccl/changefeedccl/kvfeed/scanner_test.go b/pkg/ccl/changefeedccl/kvfeed/scanner_test.go | |
| index 48b104d91c..5c7dc6238e 100644 | |
| --- a/pkg/ccl/changefeedccl/kvfeed/scanner_test.go | |
| +++ b/pkg/ccl/changefeedccl/kvfeed/scanner_test.go | |
| @@ -70,9 +70,8 @@ INSERT INTO t VALUES (1), (2), (3); | |
| Spans: []roachpb.Span{span}, | |
| Timestamp: exportTime, | |
| Knobs: TestingKnobs{ | |
| - BeforeScanRequest: func(b *kv.Batch) error { | |
| + BeforeScanRequest: func(b *kv.Batch) { | |
| b.Header.MaxSpanRequestKeys = 1 | |
| - return nil | |
| }, | |
| }, | |
| } | |
| diff --git a/pkg/ccl/changefeedccl/kvfeed/testing_knobs.go b/pkg/ccl/changefeedccl/kvfeed/testing_knobs.go | |
| index 72dc6fecce..2e8645f035 100644 | |
| --- a/pkg/ccl/changefeedccl/kvfeed/testing_knobs.go | |
| +++ b/pkg/ccl/changefeedccl/kvfeed/testing_knobs.go | |
| @@ -16,7 +16,7 @@ import ( | |
| // TestingKnobs are the testing knobs for kvfeed. | |
| type TestingKnobs struct { | |
| // BeforeScanRequest is a callback invoked before issuing Scan request. | |
| - BeforeScanRequest func(b *kv.Batch) error | |
| + BeforeScanRequest func(b *kv.Batch) | |
| OnRangeFeedValue func(kv roachpb.KeyValue) error | |
| } | |
| diff --git a/pkg/ccl/changefeedccl/metrics.go b/pkg/ccl/changefeedccl/metrics.go | |
| index e0e62c3753..3d85f035a7 100644 | |
| --- a/pkg/ccl/changefeedccl/metrics.go | |
| +++ b/pkg/ccl/changefeedccl/metrics.go | |
| @@ -10,7 +10,6 @@ package changefeedccl | |
| import ( | |
| "strings" | |
| - "sync/atomic" | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/ccl/changefeedccl/cdcutils" | |
| @@ -42,18 +41,17 @@ const defaultSLIScope = "default" | |
| // AggMetrics are aggregated metrics keeping track of aggregated changefeed performance | |
| // indicators, combined with a limited number of per-changefeed indicators. | |
| type AggMetrics struct { | |
| - EmittedMessages *aggmetric.AggCounter | |
| - EmittedBytes *aggmetric.AggCounter | |
| - FlushedBytes *aggmetric.AggCounter | |
| - BatchHistNanos *aggmetric.AggHistogram | |
| - Flushes *aggmetric.AggCounter | |
| - FlushHistNanos *aggmetric.AggHistogram | |
| - CommitLatency *aggmetric.AggHistogram | |
| - BackfillCount *aggmetric.AggGauge | |
| - BackfillPendingRanges *aggmetric.AggGauge | |
| - ErrorRetries *aggmetric.AggCounter | |
| - AdmitLatency *aggmetric.AggHistogram | |
| - RunningCount *aggmetric.AggGauge | |
| + EmittedMessages *aggmetric.AggCounter | |
| + EmittedBytes *aggmetric.AggCounter | |
| + FlushedBytes *aggmetric.AggCounter | |
| + BatchHistNanos *aggmetric.AggHistogram | |
| + Flushes *aggmetric.AggCounter | |
| + FlushHistNanos *aggmetric.AggHistogram | |
| + CommitLatency *aggmetric.AggHistogram | |
| + BackfillCount *aggmetric.AggGauge | |
| + ErrorRetries *aggmetric.AggCounter | |
| + AdmitLatency *aggmetric.AggHistogram | |
| + RunningCount *aggmetric.AggGauge | |
| // There is always at least 1 sliMetrics created for defaultSLI scope. | |
| mu struct { | |
| @@ -67,18 +65,17 @@ func (a *AggMetrics) MetricStruct() {} | |
| // sliMetrics holds all SLI related metrics aggregated into AggMetrics. | |
| type sliMetrics struct { | |
| - EmittedMessages *aggmetric.Counter | |
| - EmittedBytes *aggmetric.Counter | |
| - FlushedBytes *aggmetric.Counter | |
| - BatchHistNanos *aggmetric.Histogram | |
| - Flushes *aggmetric.Counter | |
| - FlushHistNanos *aggmetric.Histogram | |
| - CommitLatency *aggmetric.Histogram | |
| - ErrorRetries *aggmetric.Counter | |
| - AdmitLatency *aggmetric.Histogram | |
| - BackfillCount *aggmetric.Gauge | |
| - BackfillPendingRanges *aggmetric.Gauge | |
| - RunningCount *aggmetric.Gauge | |
| + EmittedMessages *aggmetric.Counter | |
| + EmittedBytes *aggmetric.Counter | |
| + FlushedBytes *aggmetric.Counter | |
| + BatchHistNanos *aggmetric.Histogram | |
| + Flushes *aggmetric.Counter | |
| + FlushHistNanos *aggmetric.Histogram | |
| + CommitLatency *aggmetric.Histogram | |
| + ErrorRetries *aggmetric.Counter | |
| + AdmitLatency *aggmetric.Histogram | |
| + BackfillCount *aggmetric.Gauge | |
| + RunningCount *aggmetric.Gauge | |
| } | |
| // sinkDoesNotCompress is a sentinel value indicating the sink | |
| @@ -152,28 +149,6 @@ func (m *sliMetrics) getBackfillCallback() func() func() { | |
| } | |
| } | |
| -// getBackfillRangeCallback returns a backfillRangeCallback that is to be called | |
| -// at the beginning of a backfill with the number of ranges that will be scanned | |
| -// and returns a two callbacks to decrement the value until all ranges have | |
| -// been emitted or clear the number completely if the backfill is cancelled. | |
| -// Note: dec() should only be called as many times as the initial value, and | |
| -// clear() should be called when there will never be another dec() call. | |
| -func (m *sliMetrics) getBackfillRangeCallback() func(int64) (func(), func()) { | |
| - return func(initial int64) (dec func(), clear func()) { | |
| - remaining := initial | |
| - m.BackfillPendingRanges.Inc(initial) | |
| - dec = func() { | |
| - m.BackfillPendingRanges.Dec(1) | |
| - atomic.AddInt64(&remaining, -1) | |
| - } | |
| - clear = func() { | |
| - m.BackfillPendingRanges.Dec(remaining) | |
| - atomic.AddInt64(&remaining, -remaining) | |
| - } | |
| - return | |
| - } | |
| -} | |
| - | |
| const ( | |
| changefeedCheckpointHistMaxLatency = 30 * time.Second | |
| changefeedBatchHistMaxLatency = 30 * time.Second | |
| @@ -297,12 +272,6 @@ func newAggregateMetrics(histogramWindow time.Duration) *AggMetrics { | |
| Measurement: "Count", | |
| Unit: metric.Unit_COUNT, | |
| } | |
| - metaChangefeedBackfillPendingRanges := metric.Metadata{ | |
| - Name: "changefeed.backfill_pending_ranges", | |
| - Help: "Number of ranges in an ongoing backfill that are yet to be fully emitted", | |
| - Measurement: "Count", | |
| - Unit: metric.Unit_COUNT, | |
| - } | |
| metaChangefeedRunning := metric.Metadata{ | |
| Name: "changefeed.running", | |
| Help: "Number of currently running changefeeds, including sinkless", | |
| @@ -328,9 +297,8 @@ func newAggregateMetrics(histogramWindow time.Duration) *AggMetrics { | |
| histogramWindow, commitLatencyMaxValue.Nanoseconds(), 1), | |
| AdmitLatency: b.Histogram(metaAdmitLatency, histogramWindow, | |
| admitLatencyMaxValue.Nanoseconds(), 1), | |
| - BackfillCount: b.Gauge(metaChangefeedBackfillCount), | |
| - BackfillPendingRanges: b.Gauge(metaChangefeedBackfillPendingRanges), | |
| - RunningCount: b.Gauge(metaChangefeedRunning), | |
| + BackfillCount: b.Gauge(metaChangefeedBackfillCount), | |
| + RunningCount: b.Gauge(metaChangefeedRunning), | |
| } | |
| a.mu.sliMetrics = make(map[string]*sliMetrics) | |
| _, err := a.getOrCreateScope(defaultSLIScope) | |
| @@ -375,18 +343,17 @@ func (a *AggMetrics) getOrCreateScope(scope string) (*sliMetrics, error) { | |
| } | |
| sm := &sliMetrics{ | |
| - EmittedMessages: a.EmittedMessages.AddChild(scope), | |
| - EmittedBytes: a.EmittedBytes.AddChild(scope), | |
| - FlushedBytes: a.FlushedBytes.AddChild(scope), | |
| - BatchHistNanos: a.BatchHistNanos.AddChild(scope), | |
| - Flushes: a.Flushes.AddChild(scope), | |
| - FlushHistNanos: a.FlushHistNanos.AddChild(scope), | |
| - CommitLatency: a.CommitLatency.AddChild(scope), | |
| - ErrorRetries: a.ErrorRetries.AddChild(scope), | |
| - AdmitLatency: a.AdmitLatency.AddChild(scope), | |
| - BackfillCount: a.BackfillCount.AddChild(scope), | |
| - BackfillPendingRanges: a.BackfillPendingRanges.AddChild(scope), | |
| - RunningCount: a.RunningCount.AddChild(scope), | |
| + EmittedMessages: a.EmittedMessages.AddChild(scope), | |
| + EmittedBytes: a.EmittedBytes.AddChild(scope), | |
| + FlushedBytes: a.FlushedBytes.AddChild(scope), | |
| + BatchHistNanos: a.BatchHistNanos.AddChild(scope), | |
| + Flushes: a.Flushes.AddChild(scope), | |
| + FlushHistNanos: a.FlushHistNanos.AddChild(scope), | |
| + CommitLatency: a.CommitLatency.AddChild(scope), | |
| + ErrorRetries: a.ErrorRetries.AddChild(scope), | |
| + AdmitLatency: a.AdmitLatency.AddChild(scope), | |
| + BackfillCount: a.BackfillCount.AddChild(scope), | |
| + RunningCount: a.RunningCount.AddChild(scope), | |
| } | |
| a.mu.sliMetrics[scope] = sm | |
| diff --git a/pkg/ccl/changefeedccl/rowfetcher_cache.go b/pkg/ccl/changefeedccl/rowfetcher_cache.go | |
| index cfdcc82990..d5fce6be70 100644 | |
| --- a/pkg/ccl/changefeedccl/rowfetcher_cache.go | |
| +++ b/pkg/ccl/changefeedccl/rowfetcher_cache.go | |
| @@ -44,9 +44,8 @@ type rowFetcherCache struct { | |
| } | |
| type cachedFetcher struct { | |
| - tableDesc catalog.TableDescriptor | |
| - fetcher row.Fetcher | |
| - familyDesc descpb.ColumnFamilyDescriptor | |
| + tableDesc catalog.TableDescriptor | |
| + fetcher row.Fetcher | |
| } | |
| var rfCacheConfig = cache.Config{ | |
| @@ -60,7 +59,6 @@ var rfCacheConfig = cache.Config{ | |
| type idVersion struct { | |
| id descpb.ID | |
| version descpb.DescriptorVersion | |
| - family descpb.FamilyID | |
| } | |
| func newRowFetcherCache( | |
| @@ -81,31 +79,24 @@ func newRowFetcherCache( | |
| func (c *rowFetcherCache) TableDescForKey( | |
| ctx context.Context, key roachpb.Key, ts hlc.Timestamp, | |
| -) (catalog.TableDescriptor, descpb.FamilyID, error) { | |
| +) (catalog.TableDescriptor, error) { | |
| var tableDesc catalog.TableDescriptor | |
| key, err := c.codec.StripTenantPrefix(key) | |
| if err != nil { | |
| - return nil, descpb.FamilyID(0), err | |
| + return nil, err | |
| } | |
| remaining, tableID, _, err := rowenc.DecodePartialTableIDIndexID(key) | |
| if err != nil { | |
| - return nil, descpb.FamilyID(0), err | |
| - } | |
| - | |
| - familyID, err := keys.DecodeFamilyKey(key) | |
| - if err != nil { | |
| - return nil, descpb.FamilyID(0), err | |
| + return nil, err | |
| } | |
| - family := descpb.FamilyID(familyID) | |
| - | |
| // Retrieve the target TableDescriptor from the lease manager. No caching | |
| // is attempted because the lease manager does its own caching. | |
| desc, err := c.leaseMgr.Acquire(ctx, ts, tableID) | |
| if err != nil { | |
| // Manager can return all kinds of errors during chaos, but based on | |
| // its usage, none of them should ever be terminal. | |
| - return nil, family, changefeedbase.MarkRetryableError(err) | |
| + return nil, changefeedbase.MarkRetryableError(err) | |
| } | |
| tableDesc = desc.Underlying().(catalog.TableDescriptor) | |
| // Immediately release the lease, since we only need it for the exact | |
| @@ -132,7 +123,7 @@ func (c *rowFetcherCache) TableDescForKey( | |
| }); err != nil { | |
| // Manager can return all kinds of errors during chaos, but based on | |
| // its usage, none of them should ever be terminal. | |
| - return nil, family, changefeedbase.MarkRetryableError(err) | |
| + return nil, changefeedbase.MarkRetryableError(err) | |
| } | |
| // Immediately release the lease, since we only need it for the exact | |
| // timestamp requested. | |
| @@ -143,12 +134,12 @@ func (c *rowFetcherCache) TableDescForKey( | |
| for skippedCols := 0; skippedCols < tableDesc.GetPrimaryIndex().NumKeyColumns(); skippedCols++ { | |
| l, err := encoding.PeekLength(remaining) | |
| if err != nil { | |
| - return nil, family, err | |
| + return nil, err | |
| } | |
| remaining = remaining[l:] | |
| } | |
| - return tableDesc, family, nil | |
| + return tableDesc, nil | |
| } | |
| func (c *rowFetcherCache) RowFetcherForTableDesc( | |
| @@ -200,62 +191,3 @@ func (c *rowFetcherCache) RowFetcherForTableDesc( | |
| c.fetchers.Add(idVer, f) | |
| return rf, nil | |
| } | |
| - | |
| -func (c *rowFetcherCache) RowFetcherForColumnFamily( | |
| - tableDesc catalog.TableDescriptor, family descpb.FamilyID, | |
| -) (*row.Fetcher, error) { | |
| - idVer := idVersion{id: tableDesc.GetID(), version: tableDesc.GetVersion(), family: family} | |
| - // Ensure that all user defined types are up to date with the cached | |
| - // version and the desired version to use the cache. It is safe to use | |
| - // UserDefinedTypeColsHaveSameVersion if we have a hit because we are | |
| - // guaranteed that the tables have the same version. Additionally, these | |
| - // fetchers are always initialized with a single tabledesc.Immutable. | |
| - // TODO (zinger): Only check types used in the relevant family. | |
| - if v, ok := c.fetchers.Get(idVer); ok { | |
| - f := v.(*cachedFetcher) | |
| - if catalog.UserDefinedTypeColsHaveSameVersion(tableDesc, f.tableDesc) { | |
| - return &f.fetcher, nil | |
| - } | |
| - } | |
| - | |
| - familyDesc, err := tableDesc.FindFamilyByID(family) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - f := &cachedFetcher{ | |
| - tableDesc: tableDesc, | |
| - familyDesc: *familyDesc, | |
| - } | |
| - rf := &f.fetcher | |
| - | |
| - var spec descpb.IndexFetchSpec | |
| - | |
| - // TODO (zinger): Make fetchColumnIDs only the family and the primary key. | |
| - // This seems to cause an error without further work but would be more efficient. | |
| - if err := rowenc.InitIndexFetchSpec( | |
| - &spec, c.codec, tableDesc, tableDesc.GetPrimaryIndex(), tableDesc.PublicColumnIDs(), | |
| - ); err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - if err := rf.Init( | |
| - context.TODO(), | |
| - false, /* reverse */ | |
| - descpb.ScanLockingStrength_FOR_NONE, | |
| - descpb.ScanLockingWaitPolicy_BLOCK, | |
| - 0, /* lockTimeout */ | |
| - &c.a, | |
| - nil, /* memMonitor */ | |
| - &spec, | |
| - ); err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - // Necessary because virtual columns are not populated. | |
| - // TODO(radu): should we stop requesting those columns from the fetcher? | |
| - rf.IgnoreUnexpectedNulls = true | |
| - | |
| - c.fetchers.Add(idVer, f) | |
| - return rf, nil | |
| -} | |
| diff --git a/pkg/ccl/changefeedccl/schemafeed/schema_feed.go b/pkg/ccl/changefeedccl/schemafeed/schema_feed.go | |
| index c4b36f5fbe..2761e653b6 100644 | |
| --- a/pkg/ccl/changefeedccl/schemafeed/schema_feed.go | |
| +++ b/pkg/ccl/changefeedccl/schemafeed/schema_feed.go | |
| @@ -81,7 +81,6 @@ func New( | |
| targets []jobspb.ChangefeedTargetSpecification, | |
| initialHighwater hlc.Timestamp, | |
| metrics *Metrics, | |
| - changefeedOpts map[string]string, | |
| ) SchemaFeed { | |
| m := &schemaFeed{ | |
| filter: schemaChangeEventFilters[events], | |
| @@ -93,7 +92,6 @@ func New( | |
| ie: cfg.SessionBoundInternalExecutorFactory(ctx, &sessiondata.SessionData{}), | |
| collectionFactory: cfg.CollectionFactory, | |
| metrics: metrics, | |
| - changefeedOpts: changefeedOpts, | |
| } | |
| m.mu.previousTableVersion = make(map[descpb.ID]catalog.TableDescriptor) | |
| m.mu.highWater = initialHighwater | |
| @@ -112,14 +110,13 @@ func New( | |
| // invariant (via `validateFn`). An error timestamp is also kept, which is the | |
| // lowest timestamp where at least one table doesn't meet the invariant. | |
| type schemaFeed struct { | |
| - filter tableEventFilter | |
| - db *kv.DB | |
| - clock *hlc.Clock | |
| - settings *cluster.Settings | |
| - targets []jobspb.ChangefeedTargetSpecification | |
| - ie sqlutil.InternalExecutor | |
| - metrics *Metrics | |
| - changefeedOpts map[string]string | |
| + filter tableEventFilter | |
| + db *kv.DB | |
| + clock *hlc.Clock | |
| + settings *cluster.Settings | |
| + targets []jobspb.ChangefeedTargetSpecification | |
| + ie sqlutil.InternalExecutor | |
| + metrics *Metrics | |
| // TODO(ajwerner): Should this live underneath the FilterFunc? | |
| // Should there be another function to decide whether to update the | |
| @@ -528,7 +525,7 @@ func (tf *schemaFeed) validateDescriptor( | |
| // manager to acquire the freshest version of the type. | |
| return tf.leaseMgr.AcquireFreshestFromStore(ctx, desc.GetID()) | |
| case catalog.TableDescriptor: | |
| - if err := changefeedbase.ValidateTable(tf.targets, desc, tf.changefeedOpts); err != nil { | |
| + if err := changefeedbase.ValidateTable(tf.targets, desc); err != nil { | |
| return err | |
| } | |
| log.VEventf(ctx, 1, "validate %v", formatDesc(desc)) | |
| diff --git a/pkg/ccl/changefeedccl/sink_kafka.go b/pkg/ccl/changefeedccl/sink_kafka.go | |
| index 95f827bdd7..273578cfcb 100644 | |
| --- a/pkg/ccl/changefeedccl/sink_kafka.go | |
| +++ b/pkg/ccl/changefeedccl/sink_kafka.go | |
| @@ -360,11 +360,6 @@ func (s *kafkaSink) workerLoop() { | |
| ackMsg = m | |
| case err := <-s.producer.Errors(): | |
| ackMsg, ackError = err.Msg, err.Err | |
| - if ackError != nil { | |
| - ackError = errors.Wrapf(ackError, | |
| - "while sending message with key=%s, size=%d", | |
| - err.Msg.Key, err.Msg.Key.Length()+err.Msg.Value.Length()) | |
| - } | |
| } | |
| if m, ok := ackMsg.Metadata.(messageMetadata); ok { | |
| diff --git a/pkg/ccl/cliccl/debug.go b/pkg/ccl/cliccl/debug.go | |
| index 138b4bdf39..81bcbfd5a4 100644 | |
| --- a/pkg/ccl/cliccl/debug.go | |
| +++ b/pkg/ccl/cliccl/debug.go | |
| @@ -160,7 +160,7 @@ func runEncryptionStatus(cmd *cobra.Command, args []string) error { | |
| dir := args[0] | |
| - db, err := cli.OpenExistingStore(dir, stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := cli.OpenExistingStore(dir, stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| diff --git a/pkg/ccl/kvccl/kvtenantccl/BUILD.bazel b/pkg/ccl/kvccl/kvtenantccl/BUILD.bazel | |
| index a9c2164988..16ec8993c2 100644 | |
| --- a/pkg/ccl/kvccl/kvtenantccl/BUILD.bazel | |
| +++ b/pkg/ccl/kvccl/kvtenantccl/BUILD.bazel | |
| @@ -48,7 +48,6 @@ go_test( | |
| "tenant_upgrade_test.go", | |
| ], | |
| embed = [":kvtenantccl"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/ccl/utilccl", | |
| @@ -85,7 +84,6 @@ go_test( | |
| "//pkg/util/stop", | |
| "//pkg/util/tracing", | |
| "//pkg/util/uuid", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| "@com_github_stretchr_testify//assert", | |
| "@com_github_stretchr_testify//require", | |
| "@org_golang_google_grpc//codes", | |
| diff --git a/pkg/ccl/kvccl/kvtenantccl/connector.go b/pkg/ccl/kvccl/kvtenantccl/connector.go | |
| index 1b28cad4f8..ddce7fd207 100644 | |
| --- a/pkg/ccl/kvccl/kvtenantccl/connector.go | |
| +++ b/pkg/ccl/kvccl/kvtenantccl/connector.go | |
| @@ -120,11 +120,6 @@ var _ config.SystemConfigProvider = (*Connector)(nil) | |
| // multi-region primitives. | |
| var _ serverpb.RegionsServer = (*Connector)(nil) | |
| -// Connector is capable of finding debug information about the current | |
| -// tenant within the cluster. This is necessary for things such as | |
| -// debug zip and range reports. | |
| -var _ serverpb.TenantStatusServer = (*Connector)(nil) | |
| - | |
| // Connector is capable of accessing span configurations for secondary tenants. | |
| var _ spanconfig.KVAccessor = (*Connector)(nil) | |
| @@ -433,21 +428,6 @@ func (c *Connector) Regions( | |
| return resp, nil | |
| } | |
| -// TenantRanges implements the serverpb.TenantStatusServer interface | |
| -func (c *Connector) TenantRanges( | |
| - ctx context.Context, req *serverpb.TenantRangesRequest, | |
| -) (resp *serverpb.TenantRangesResponse, _ error) { | |
| - if err := c.withClient(ctx, func(ctx context.Context, c *client) error { | |
| - var err error | |
| - resp, err = c.TenantRanges(ctx, req) | |
| - return err | |
| - }); err != nil { | |
| - return nil, err | |
| - } | |
| - | |
| - return resp, nil | |
| -} | |
| - | |
| // FirstRange implements the kvcoord.RangeDescriptorDB interface. | |
| func (c *Connector) FirstRange() (*roachpb.RangeDescriptor, error) { | |
| return nil, status.Error(codes.Unauthenticated, "kvtenant.Proxy does not have access to FirstRange") | |
| diff --git a/pkg/ccl/kvccl/kvtenantccl/tenant_trace_test.go b/pkg/ccl/kvccl/kvtenantccl/tenant_trace_test.go | |
| index dcb8819ce3..4399ccdba4 100644 | |
| --- a/pkg/ccl/kvccl/kvtenantccl/tenant_trace_test.go | |
| +++ b/pkg/ccl/kvccl/kvtenantccl/tenant_trace_test.go | |
| @@ -26,7 +26,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| - "github.com/cockroachdb/redact" | |
| "github.com/stretchr/testify/require" | |
| ) | |
| @@ -58,7 +57,7 @@ func testTenantTracesAreRedactedImpl(t *testing.T, redactable bool) { | |
| EvalKnobs: kvserverbase.BatchEvalTestingKnobs{ | |
| TestingEvalFilter: func(args kvserverbase.FilterArgs) *roachpb.Error { | |
| log.Eventf(args.Ctx, "%v", sensitiveString) | |
| - log.Eventf(args.Ctx, "%v", redact.Safe(visibleString)) | |
| + log.Eventf(args.Ctx, "%v", log.Safe(visibleString)) | |
| return nil | |
| }, | |
| }, | |
| diff --git a/pkg/ccl/logictestccl/testdata/logic_test/crdb_internal_tenant b/pkg/ccl/logictestccl/testdata/logic_test/crdb_internal_tenant | |
| index 0c28ebb0b3..c4477aa8a6 100644 | |
| --- a/pkg/ccl/logictestccl/testdata/logic_test/crdb_internal_tenant | |
| +++ b/pkg/ccl/logictestccl/testdata/logic_test/crdb_internal_tenant | |
| @@ -97,7 +97,6 @@ crdb_internal table_indexes table NULL NULL NULL | |
| crdb_internal table_row_statistics table NULL NULL NULL | |
| crdb_internal tables table NULL NULL NULL | |
| crdb_internal tenant_usage_details view NULL NULL NULL | |
| -crdb_internal transaction_contention_events table NULL NULL NULL | |
| crdb_internal transaction_statistics view NULL NULL NULL | |
| crdb_internal zones table NULL NULL NULL | |
| diff --git a/pkg/ccl/logictestccl/testdata/logic_test/partitioning_hash_sharded_index b/pkg/ccl/logictestccl/testdata/logic_test/partitioning_hash_sharded_index | |
| index 03b696cc89..b7706cbc31 100644 | |
| --- a/pkg/ccl/logictestccl/testdata/logic_test/partitioning_hash_sharded_index | |
| +++ b/pkg/ccl/logictestccl/testdata/logic_test/partitioning_hash_sharded_index | |
| @@ -281,4 +281,4 @@ t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/3 /Tabl | |
| t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/4 /Table/116/2/"seattle"/5 | |
| t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/5 /Table/116/2/"seattle"/6 | |
| t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/6 /Table/116/2/"seattle"/7 | |
| -t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/7 /Table/116/3/"new york"/0 | |
| +t_presplit 116 t_presplit_idx_member_id /Table/116/2/"seattle"/7 /Max | |
| diff --git a/pkg/ccl/logictestccl/testdata/logic_test/regional_by_row_hash_sharded_index_query_plan b/pkg/ccl/logictestccl/testdata/logic_test/regional_by_row_hash_sharded_index_query_plan | |
| index 585df35668..76e0db7d72 100644 | |
| --- a/pkg/ccl/logictestccl/testdata/logic_test/regional_by_row_hash_sharded_index_query_plan | |
| +++ b/pkg/ccl/logictestccl/testdata/logic_test/regional_by_row_hash_sharded_index_query_plan | |
| @@ -36,7 +36,7 @@ CREATE TABLE t_child_regional ( | |
| # TODO (mgartner): there is a lookup join with lookup condition checking every | |
| # single shard. This is unnecessary and could be improved by having the shard | |
| # number calculated instead of looking at all possible shards. | |
| -query T retry | |
| +query T | |
| EXPLAIN (VERBOSE) INSERT INTO t_child VALUES (123, 321); | |
| ---- | |
| distribution: local | |
| diff --git a/pkg/ccl/schemachangerccl/BUILD.bazel b/pkg/ccl/schemachangerccl/BUILD.bazel | |
| index bd5a4da608..956253970d 100644 | |
| --- a/pkg/ccl/schemachangerccl/BUILD.bazel | |
| +++ b/pkg/ccl/schemachangerccl/BUILD.bazel | |
| @@ -11,7 +11,6 @@ go_test( | |
| "//pkg/sql/schemachanger:testdata", | |
| ], | |
| embed = [":schemachangerccl"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/build/bazel", | |
| diff --git a/pkg/ccl/serverccl/statusccl/BUILD.bazel b/pkg/ccl/serverccl/statusccl/BUILD.bazel | |
| index 034fefe8d6..ded721723a 100644 | |
| --- a/pkg/ccl/serverccl/statusccl/BUILD.bazel | |
| +++ b/pkg/ccl/serverccl/statusccl/BUILD.bazel | |
| @@ -10,8 +10,6 @@ go_library( | |
| "//pkg/roachpb", | |
| "//pkg/security", | |
| "//pkg/server/serverpb", | |
| - "//pkg/sql", | |
| - "//pkg/sql/contention", | |
| "//pkg/sql/pgwire", | |
| "//pkg/sql/sqlstats/persistedsqlstats", | |
| "//pkg/sql/tests", | |
| @@ -36,7 +34,6 @@ go_test( | |
| "//pkg/ccl", | |
| "//pkg/ccl/kvccl", | |
| "//pkg/ccl/utilccl", | |
| - "//pkg/keys", | |
| "//pkg/roachpb", | |
| "//pkg/rpc", | |
| "//pkg/security", | |
| diff --git a/pkg/ccl/serverccl/statusccl/tenant_status_test.go b/pkg/ccl/serverccl/statusccl/tenant_status_test.go | |
| index b615a2265c..2633187ca8 100644 | |
| --- a/pkg/ccl/serverccl/statusccl/tenant_status_test.go | |
| +++ b/pkg/ccl/serverccl/statusccl/tenant_status_test.go | |
| @@ -22,7 +22,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/base" | |
| _ "github.com/cockroachdb/cockroach/pkg/ccl/kvccl" | |
| - "github.com/cockroachdb/cockroach/pkg/keys" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/security" | |
| "github.com/cockroachdb/cockroach/pkg/server/serverpb" | |
| @@ -97,10 +96,6 @@ func TestTenantStatusAPI(t *testing.T) { | |
| t.Run("txn_id_resolution", func(t *testing.T) { | |
| testTxnIDResolutionRPC(ctx, t, testHelper) | |
| }) | |
| - | |
| - t.Run("tenant_ranges", func(t *testing.T) { | |
| - testTenantRangesRPC(ctx, t, testHelper) | |
| - }) | |
| } | |
| func TestTenantCannotSeeNonTenantStats(t *testing.T) { | |
| @@ -718,26 +713,6 @@ SET TRACING=off; | |
| t.Errorf("did not expect contention event in controlled cluster, but it was found") | |
| } | |
| } | |
| - | |
| - testutils.SucceedsWithin(t, func() error { | |
| - err = testHelper.testCluster().tenantContentionRegistry(1).FlushEventsForTest(ctx) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - | |
| - resp := &serverpb.TransactionContentionEventsResponse{} | |
| - testHelper. | |
| - testCluster(). | |
| - tenantHTTPClient(t, 1). | |
| - GetJSON("/_status/transactioncontentionevents", resp) | |
| - | |
| - if len(resp.Events) == 0 { | |
| - return errors.New("expected transaction contention events being populated, " + | |
| - "but it is not") | |
| - } | |
| - | |
| - return nil | |
| - }, 5*time.Second) | |
| } | |
| func testIndexUsageForTenants(t *testing.T, testHelper *tenantTestHelper) { | |
| @@ -949,7 +924,7 @@ func testTxnIDResolutionRPC(ctx context.Context, t *testing.T, helper *tenantTes | |
| "expected a valid txnID, but %+v is found", result) | |
| sqlConn.Exec(t, "COMMIT") | |
| - testutils.SucceedsSoon(t, func() error { | |
| + testutils.SucceedsWithin(t, func() error { | |
| resp, err := status.TxnIDResolution(ctx, &serverpb.TxnIDResolutionRequest{ | |
| CoordinatorID: strconv.Itoa(int(coordinatorNodeID)), | |
| TxnIDs: []uuid.UUID{txnID}, | |
| @@ -964,7 +939,7 @@ func testTxnIDResolutionRPC(ctx context.Context, t *testing.T, helper *tenantTes | |
| "was not", txnID.String(), coordinatorNodeID) | |
| require.NotEqual(t, roachpb.InvalidTransactionFingerprintID, resp.ResolvedTxnIDs[0].TxnFingerprintID) | |
| return nil | |
| - }) | |
| + }, 3*time.Second) | |
| } | |
| t.Run("regular_cluster", func(t *testing.T) { | |
| @@ -983,60 +958,3 @@ func testTxnIDResolutionRPC(ctx context.Context, t *testing.T, helper *tenantTes | |
| run(sqlConn, status, 1 /* coordinatorNodeID */) | |
| }) | |
| } | |
| - | |
| -func testTenantRangesRPC(_ context.Context, t *testing.T, helper *tenantTestHelper) { | |
| - tenantA := helper.testCluster().tenant(0).tenant.TenantStatusServer().(serverpb.TenantStatusServer) | |
| - keyPrefixForA := keys.MakeTenantPrefix(helper.testCluster().tenant(0).tenant.RPCContext().TenantID) | |
| - keyPrefixEndForA := keyPrefixForA.PrefixEnd() | |
| - | |
| - tenantB := helper.controlCluster().tenant(0).tenant.TenantStatusServer().(serverpb.TenantStatusServer) | |
| - keyPrefixForB := keys.MakeTenantPrefix(helper.controlCluster().tenant(0).tenant.RPCContext().TenantID) | |
| - keyPrefixEndForB := keyPrefixForB.PrefixEnd() | |
| - | |
| - resp, err := tenantA.TenantRanges(context.Background(), &serverpb.TenantRangesRequest{}) | |
| - require.NoError(t, err) | |
| - require.NotEmpty(t, resp.RangesByLocality) | |
| - for localityKey, rangeList := range resp.RangesByLocality { | |
| - require.NotEmpty(t, localityKey) | |
| - for _, r := range rangeList.Ranges { | |
| - assertStartKeyInRange(t, r.Span.StartKey, keyPrefixForA) | |
| - assertEndKeyInRange(t, r.Span.EndKey, keyPrefixForA, keyPrefixEndForA) | |
| - } | |
| - } | |
| - | |
| - resp, err = tenantB.TenantRanges(context.Background(), &serverpb.TenantRangesRequest{}) | |
| - require.NoError(t, err) | |
| - require.NotEmpty(t, resp.RangesByLocality) | |
| - for localityKey, rangeList := range resp.RangesByLocality { | |
| - require.NotEmpty(t, localityKey) | |
| - for _, r := range rangeList.Ranges { | |
| - assertStartKeyInRange(t, r.Span.StartKey, keyPrefixForB) | |
| - assertEndKeyInRange(t, r.Span.EndKey, keyPrefixForB, keyPrefixEndForB) | |
| - } | |
| - } | |
| -} | |
| - | |
| -// assertStartKeyInRange compares the pretty printed startKey with the provided | |
| -// tenantPrefix key, ensuring that the startKey starts with the tenantPrefix. | |
| -func assertStartKeyInRange(t *testing.T, startKey string, tenantPrefix roachpb.Key) { | |
| - require.Truef(t, strings.Index(startKey, tenantPrefix.String()) == 0, | |
| - fmt.Sprintf("start key %s is outside of the tenant's keyspace (prefix: %v)", | |
| - startKey, tenantPrefix.String())) | |
| -} | |
| - | |
| -// assertEndKeyInRange compares the pretty printed endKey with the provided | |
| -// tenantPrefix and tenantPrefixEnd keys. Ensures that the key starts with | |
| -// either the tenantPrefix, or the tenantPrefixEnd (valid as end keys are | |
| -// exclusive). | |
| -func assertEndKeyInRange( | |
| - t *testing.T, endKey string, tenantPrefix roachpb.Key, tenantPrefixEnd roachpb.Key, | |
| -) { | |
| - require.Truef(t, | |
| - strings.Index(endKey, tenantPrefix.String()) == 0 || | |
| - strings.Index(endKey, tenantPrefixEnd.String()) == 0 || | |
| - // Possible if the tenant's ranges fall at the end of the entire keyspace | |
| - // range within the cluster. | |
| - endKey == "/Max", | |
| - fmt.Sprintf("end key %s is outside of the tenant's keyspace (prefix: %v, prefixEnd: %v)", | |
| - endKey, tenantPrefix.String(), tenantPrefixEnd.String())) | |
| -} | |
| diff --git a/pkg/ccl/serverccl/statusccl/tenant_test_utils.go b/pkg/ccl/serverccl/statusccl/tenant_test_utils.go | |
| index 4156a271e2..91c5a15acc 100644 | |
| --- a/pkg/ccl/serverccl/statusccl/tenant_test_utils.go | |
| +++ b/pkg/ccl/serverccl/statusccl/tenant_test_utils.go | |
| @@ -19,8 +19,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/security" | |
| "github.com/cockroachdb/cockroach/pkg/server/serverpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/contention" | |
| "github.com/cockroachdb/cockroach/pkg/sql/pgwire" | |
| "github.com/cockroachdb/cockroach/pkg/sql/sqlstats/persistedsqlstats" | |
| "github.com/cockroachdb/cockroach/pkg/sql/tests" | |
| @@ -39,12 +37,11 @@ type serverIdx int | |
| const randomServer serverIdx = -1 | |
| type testTenant struct { | |
| - tenant serverutils.TestTenantInterface | |
| - tenantConn *gosql.DB | |
| - tenantDB *sqlutils.SQLRunner | |
| - tenantStatus serverpb.SQLStatusServer | |
| - tenantSQLStats *persistedsqlstats.PersistedSQLStats | |
| - tenantContentionRegistry *contention.Registry | |
| + tenant serverutils.TestTenantInterface | |
| + tenantConn *gosql.DB | |
| + tenantDB *sqlutils.SQLRunner | |
| + tenantStatus serverpb.SQLStatusServer | |
| + tenantSQLStats *persistedsqlstats.PersistedSQLStats | |
| } | |
| func newTestTenant( | |
| @@ -65,15 +62,13 @@ func newTestTenant( | |
| status := tenant.StatusServer().(serverpb.SQLStatusServer) | |
| sqlStats := tenant.PGServer().(*pgwire.Server).SQLServer. | |
| GetSQLStatsProvider().(*persistedsqlstats.PersistedSQLStats) | |
| - contentionRegistry := tenant.ExecutorConfig().(sql.ExecutorConfig).ContentionRegistry | |
| return &testTenant{ | |
| - tenant: tenant, | |
| - tenantConn: tenantConn, | |
| - tenantDB: sqlDB, | |
| - tenantStatus: status, | |
| - tenantSQLStats: sqlStats, | |
| - tenantContentionRegistry: contentionRegistry, | |
| + tenant: tenant, | |
| + tenantConn: tenantConn, | |
| + tenantDB: sqlDB, | |
| + tenantStatus: status, | |
| + tenantSQLStats: sqlStats, | |
| } | |
| } | |
| @@ -177,10 +172,6 @@ func (c tenantCluster) tenantStatusSrv(idx serverIdx) serverpb.SQLStatusServer { | |
| return c.tenant(idx).tenantStatus | |
| } | |
| -func (c tenantCluster) tenantContentionRegistry(idx serverIdx) *contention.Registry { | |
| - return c.tenant(idx).tenantContentionRegistry | |
| -} | |
| - | |
| func (c tenantCluster) cleanup(t *testing.T) { | |
| for _, tenant := range c { | |
| tenant.cleanup(t) | |
| diff --git a/pkg/ccl/sqlproxyccl/BUILD.bazel b/pkg/ccl/sqlproxyccl/BUILD.bazel | |
| index 69a8d24b3d..8eec3dd6ba 100644 | |
| --- a/pkg/ccl/sqlproxyccl/BUILD.bazel | |
| +++ b/pkg/ccl/sqlproxyccl/BUILD.bazel | |
| @@ -6,7 +6,6 @@ go_library( | |
| srcs = [ | |
| "authentication.go", | |
| "backend_dialer.go", | |
| - "conn_migration.go", | |
| "connector.go", | |
| "error.go", | |
| "forwarder.go", | |
| @@ -29,7 +28,6 @@ go_library( | |
| "//pkg/security/certmgr", | |
| "//pkg/sql/pgwire", | |
| "//pkg/sql/pgwire/pgcode", | |
| - "//pkg/sql/pgwire/pgwirebase", | |
| "//pkg/util/contextutil", | |
| "//pkg/util/grpcutil", | |
| "//pkg/util/httputil", | |
| @@ -54,7 +52,6 @@ go_test( | |
| size = "small", | |
| srcs = [ | |
| "authentication_test.go", | |
| - "conn_migration_test.go", | |
| "connector_test.go", | |
| "forwarder_test.go", | |
| "frontend_admitter_test.go", | |
| @@ -68,7 +65,6 @@ go_test( | |
| "//pkg/base", | |
| "//pkg/ccl/kvccl/kvtenantccl", | |
| "//pkg/ccl/sqlproxyccl/denylist", | |
| - "//pkg/ccl/sqlproxyccl/interceptor", | |
| "//pkg/ccl/sqlproxyccl/tenantdirsvr", | |
| "//pkg/ccl/sqlproxyccl/throttler", | |
| "//pkg/ccl/utilccl", | |
| diff --git a/pkg/ccl/sqlproxyccl/authentication.go b/pkg/ccl/sqlproxyccl/authentication.go | |
| index f1635fd43f..e9e8f32604 100644 | |
| --- a/pkg/ccl/sqlproxyccl/authentication.go | |
| +++ b/pkg/ccl/sqlproxyccl/authentication.go | |
| @@ -153,19 +153,19 @@ var authenticate = func(clientConn, crdbConn net.Conn, throttleHook func(throttl | |
| // we should merge them back in the future. Instead of having the writer as the | |
| // other end, the writer should be the same connection. That way, a | |
| // sqlproxyccl.Conn can be used to read-from, or write-to the same component. | |
| -var readTokenAuthResult = func(conn net.Conn) error { | |
| +var readTokenAuthResult = func(serverConn net.Conn) error { | |
| // This interceptor is discarded once this function returns. Just like | |
| - // pgproto3.NewFrontend, this serverConn object has an internal buffer. | |
| + // pgproto3.NewFrontend, this interceptor has an internal buffer. | |
| // Discarding the buffer is fine since there won't be any other messages | |
| // from the server once we receive the ReadyForQuery message because the | |
| // caller (i.e. proxy) does not forward client messages until then. | |
| - serverConn := interceptor.NewFrontendConn(conn) | |
| + serverInterceptor := interceptor.NewFrontendInterceptor(serverConn) | |
| // The auth step should require only a few back and forths so 20 iterations | |
| // should be enough. | |
| var i int | |
| for ; i < 20; i++ { | |
| - backendMsg, err := serverConn.ReadMsg() | |
| + backendMsg, err := serverInterceptor.ReadMsg() | |
| if err != nil { | |
| return newErrorf(codeBackendReadFailed, "unable to receive message from backend: %v", err) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/conn_migration.go b/pkg/ccl/sqlproxyccl/conn_migration.go | |
| deleted file mode 100644 | |
| index 3224e75f31..0000000000 | |
| --- a/pkg/ccl/sqlproxyccl/conn_migration.go | |
| +++ /dev/null | |
| @@ -1,371 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Licensed as a CockroachDB Enterprise file under the Cockroach Community | |
| -// License (the "License"); you may not use this file except in compliance with | |
| -// the License. You may obtain a copy of the License at | |
| -// | |
| -// https://github.com/cockroachdb/cockroach/blob/master/licenses/CCL.txt | |
| - | |
| -package sqlproxyccl | |
| - | |
| -import ( | |
| - "context" | |
| - "encoding/json" | |
| - "fmt" | |
| - "io" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgwirebase" | |
| - "github.com/cockroachdb/errors" | |
| - pgproto3 "github.com/jackc/pgproto3/v2" | |
| -) | |
| - | |
| -// runShowTransferState sends a SHOW TRANSFER STATE query with the input | |
| -// transferKey to the given writer. The transferKey will be used to uniquely | |
| -// identify the request when parsing the response messages in | |
| -// waitForShowTransferState. | |
| -// | |
| -// Unlike runAndWaitForDeserializeSession, we split the SHOW TRANSFER STATE | |
| -// operation into `run` and `wait` since they both will be invoked in different | |
| -// goroutines. If we combined them, we'll have to wait for at least one of the | |
| -// goroutines to pause, which can introduce a latency of about 1-2s per transfer | |
| -// while waiting for Read in readTimeoutConn to be unblocked. | |
| -func runShowTransferState(w io.Writer, transferKey string) error { | |
| - return writeQuery(w, "SHOW TRANSFER STATE WITH '%s'", transferKey) | |
| -} | |
| - | |
| -// waitForShowTransferState retrieves the transfer state from the SQL pod | |
| -// through SHOW TRANSFER STATE WITH 'key'. It is assumed that the last message | |
| -// from the server was ReadyForQuery, so the server is ready to accept a query. | |
| -// Since ReadyForQuery may be for a previous pipelined query, this handles the | |
| -// forwarding of messages back to the client in case we don't see our state yet. | |
| -// | |
| -// WARNING: When using this, we assume that no other goroutines are using both | |
| -// serverConn and clientConn. In the context of a transfer, the response | |
| -// processor must be blocked to avoid concurrent reads from serverConn. | |
| -var waitForShowTransferState = func( | |
| - ctx context.Context, | |
| - serverConn *interceptor.FrontendConn, | |
| - clientConn io.Writer, | |
| - transferKey string, | |
| -) (transferErr string, state string, revivalToken string, retErr error) { | |
| - // Wait for a response that looks like the following: | |
| - // | |
| - // error | session_state_base64 | session_revival_token_base64 | transfer_key | |
| - // --------+----------------------+------------------------------+--------------- | |
| - // NULL | .................... | ............................ | <transferKey> | |
| - // (1 row) | |
| - // | |
| - // Postgres messages always come in the following order for the | |
| - // SHOW TRANSFER STATE WITH '<transferKey>' query: | |
| - // 1. RowDescription | |
| - // 2. DataRow | |
| - // 3. CommandComplete | |
| - // 4. ReadyForQuery | |
| - | |
| - // 1. Wait for the relevant RowDescription. | |
| - if err := waitForSmallRowDescription( | |
| - ctx, | |
| - serverConn, | |
| - clientConn, | |
| - func(msg *pgproto3.RowDescription) bool { | |
| - // Do we have the right number of columns? | |
| - if len(msg.Fields) != 4 { | |
| - return false | |
| - } | |
| - // Do the names of the columns match? | |
| - var transferStateCols = []string{ | |
| - "error", | |
| - "session_state_base64", | |
| - "session_revival_token_base64", | |
| - "transfer_key", | |
| - } | |
| - for i, col := range transferStateCols { | |
| - if string(msg.Fields[i].Name) != col { | |
| - return false | |
| - } | |
| - } | |
| - return true | |
| - }, | |
| - ); err != nil { | |
| - return "", "", "", errors.Wrap(err, "waiting for RowDescription") | |
| - } | |
| - | |
| - // 2. Read DataRow. | |
| - if err := expectDataRow(ctx, serverConn, func(msg *pgproto3.DataRow) bool { | |
| - // This has to be 4 since we validated RowDescription earlier. | |
| - if len(msg.Values) != 4 { | |
| - return false | |
| - } | |
| - | |
| - // Validate transfer key. It is possible that the end-user uses the SHOW | |
| - // TRANSFER STATE WITH 'transfer_key' statement, but that isn't designed | |
| - // for external usage, so it is fine to just terminate here if the | |
| - // transfer key does not match. | |
| - if string(msg.Values[3]) != transferKey { | |
| - return false | |
| - } | |
| - | |
| - // NOTE: We have to cast to string and copy here since the slice | |
| - // referenced in msg will no longer be valid once we read the next pgwire | |
| - // message. | |
| - transferErr, state, revivalToken = string(msg.Values[0]), string(msg.Values[1]), string(msg.Values[2]) | |
| - return true | |
| - }); err != nil { | |
| - return "", "", "", errors.Wrap(err, "expecting DataRow") | |
| - } | |
| - | |
| - // 3. Read CommandComplete. | |
| - if err := expectCommandComplete(ctx, serverConn, "SHOW TRANSFER STATE 1"); err != nil { | |
| - return "", "", "", errors.Wrap(err, "expecting CommandComplete") | |
| - } | |
| - | |
| - // 4. Read ReadyForQuery. | |
| - if err := expectReadyForQuery(ctx, serverConn); err != nil { | |
| - return "", "", "", errors.Wrap(err, "expecting ReadyForQuery") | |
| - } | |
| - | |
| - return transferErr, state, revivalToken, nil | |
| -} | |
| - | |
| -// runAndWaitForDeserializeSession deserializes state into the SQL pod through | |
| -// crdb_internal.deserialize_session. It is assumed that the last message from | |
| -// the server was ReadyForQuery, so the server is ready to accept a query. | |
| -// | |
| -// This is meant to be used with a new connection, and nothing needs to be | |
| -// forwarded back to the client. | |
| -// | |
| -// WARNING: When using this, we assume that no other goroutines are using both | |
| -// serverConn and clientConn. | |
| -var runAndWaitForDeserializeSession = func( | |
| - ctx context.Context, serverConn *interceptor.FrontendConn, state string, | |
| -) error { | |
| - // Send deserialization query. | |
| - if err := writeQuery(serverConn, | |
| - "SELECT crdb_internal.deserialize_session(decode('%s', 'base64'))", state); err != nil { | |
| - return err | |
| - } | |
| - | |
| - // Wait for a response that looks like the following: | |
| - // | |
| - // crdb_internal.deserialize_session | |
| - // ------------------------------------- | |
| - // true | |
| - // (1 row) | |
| - // | |
| - // Postgres messages always come in the following order for the | |
| - // deserialize_session query: | |
| - // 1. RowDescription | |
| - // 2. DataRow | |
| - // 3. CommandComplete | |
| - // 4. ReadyForQuery | |
| - | |
| - // 1. Read RowDescription. We reuse waitFor here for convenience when we are | |
| - // really expecting instead. This is fine because we only deserialize a | |
| - // session for a new connection which hasn't been handed off to the user, | |
| - // so we can guarantee that there won't be pipelined queries. | |
| - if err := waitForSmallRowDescription( | |
| - ctx, | |
| - serverConn, | |
| - &errWriter{}, | |
| - func(msg *pgproto3.RowDescription) bool { | |
| - return len(msg.Fields) == 1 && | |
| - string(msg.Fields[0].Name) == "crdb_internal.deserialize_session" | |
| - }, | |
| - ); err != nil { | |
| - return errors.Wrap(err, "expecting RowDescription") | |
| - } | |
| - | |
| - // 2. Read DataRow. | |
| - if err := expectDataRow(ctx, serverConn, func(msg *pgproto3.DataRow) bool { | |
| - return len(msg.Values) == 1 && string(msg.Values[0]) == "t" | |
| - }); err != nil { | |
| - return errors.Wrap(err, "expecting DataRow") | |
| - } | |
| - | |
| - // 3. Read CommandComplete. | |
| - if err := expectCommandComplete(ctx, serverConn, "SELECT 1"); err != nil { | |
| - return errors.Wrap(err, "expecting CommandComplete") | |
| - } | |
| - | |
| - // 4. Read ReadyForQuery. | |
| - if err := expectReadyForQuery(ctx, serverConn); err != nil { | |
| - return errors.Wrap(err, "expecting ReadyForQuery") | |
| - } | |
| - | |
| - return nil | |
| -} | |
| - | |
| -// writeQuery writes a SimpleQuery to the given writer w. | |
| -func writeQuery(w io.Writer, format string, a ...interface{}) error { | |
| - query := &pgproto3.Query{String: fmt.Sprintf(format, a...)} | |
| - _, err := w.Write(query.Encode(nil)) | |
| - return err | |
| -} | |
| - | |
| -// waitForSmallRowDescription waits until the next message from serverConn | |
| -// is a *small* RowDescription message (i.e. within 4K bytes), and one that | |
| -// passes matchFn. When that happens, this returns nil. | |
| -// | |
| -// For all other messages (i.e. non RowDescription or large messages), they will | |
| -// be forwarded to clientConn. One exception to this would be the ErrorResponse | |
| -// message, which will result in an error since we're in an ambiguous state. | |
| -// The ErrorResponse message may be for a pipelined query, or the RowDescription | |
| -// message that we're waiting. | |
| -func waitForSmallRowDescription( | |
| - ctx context.Context, | |
| - serverConn *interceptor.FrontendConn, | |
| - clientConn io.Writer, | |
| - matchFn func(*pgproto3.RowDescription) bool, | |
| -) error { | |
| - // Since we're waiting for the first message that matches the given | |
| - // condition, we're going to loop here until we find one. | |
| - for { | |
| - if ctx.Err() != nil { | |
| - return ctx.Err() | |
| - } | |
| - | |
| - typ, size, err := serverConn.PeekMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "peeking message") | |
| - } | |
| - | |
| - // We don't know if the ErrorResponse is for the expected RowDescription | |
| - // or a previous pipelined query, so return an error. | |
| - if typ == pgwirebase.ServerMsgErrorResponse { | |
| - // Error messages are small, so read for debugging purposes. | |
| - msg, err := serverConn.ReadMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "ambiguous ErrorResponse") | |
| - } | |
| - return errors.Newf("ambiguous ErrorResponse: %v", jsonOrRaw(msg)) | |
| - } | |
| - | |
| - // Messages are intended for the client in two cases: | |
| - // 1. We have not seen a RowDescription message yet | |
| - // 2. Message was too large. This function only expects a few columns. | |
| - // | |
| - // This is mostly an optimization, and there's no point reading such | |
| - // messages into memory, so we'll just forward them back to the client | |
| - // right away. | |
| - const maxSmallMsgSize = 1 << 12 // 4KB | |
| - if typ != pgwirebase.ServerMsgRowDescription || size > maxSmallMsgSize { | |
| - if _, err := serverConn.ForwardMsg(clientConn); err != nil { | |
| - return errors.Wrap(err, "forwarding message") | |
| - } | |
| - continue | |
| - } | |
| - | |
| - msg, err := serverConn.ReadMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "reading RowDescription") | |
| - } | |
| - | |
| - pgMsg, ok := msg.(*pgproto3.RowDescription) | |
| - if !ok { | |
| - // This case will not occur since have validated the type earlier. | |
| - return errors.Newf("unexpected message: %v", jsonOrRaw(msg)) | |
| - } | |
| - | |
| - // We have found our desired RowDescription. | |
| - if matchFn(pgMsg) { | |
| - return nil | |
| - } | |
| - | |
| - // Matching fails, so forward the message back to the client, and | |
| - // continue searching. | |
| - if _, err := clientConn.Write(msg.Encode(nil)); err != nil { | |
| - return errors.Wrap(err, "writing message") | |
| - } | |
| - } | |
| -} | |
| - | |
| -// expectDataRow expects that the next message from serverConn is a DataRow | |
| -// message. If the next message is a DataRow message, validateFn will be called | |
| -// to validate the contents. This function will return an error if we don't see | |
| -// a DataRow message or the validation failed. | |
| -// | |
| -// WARNING: Use this with care since this reads the entire message into memory. | |
| -// Unlike the other expectX methods, DataRow messages may be large, and this | |
| -// does not check for that. We are currently only using this for the SHOW | |
| -// TRANSFER and crdb_internal.deserialize_session() statements, and they both | |
| -// have been vetted. The former's size will be guarded behind a cluster setting, | |
| -// whereas for the latter, the response is expected to be small. | |
| -func expectDataRow( | |
| - ctx context.Context, | |
| - serverConn *interceptor.FrontendConn, | |
| - validateFn func(*pgproto3.DataRow) bool, | |
| -) error { | |
| - if ctx.Err() != nil { | |
| - return ctx.Err() | |
| - } | |
| - msg, err := serverConn.ReadMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "reading message") | |
| - } | |
| - pgMsg, ok := msg.(*pgproto3.DataRow) | |
| - if !ok { | |
| - return errors.Newf("unexpected message: %v", jsonOrRaw(msg)) | |
| - } | |
| - if !validateFn(pgMsg) { | |
| - return errors.Newf("validation failed for message: %v", jsonOrRaw(msg)) | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -// expectCommandComplete expects that the next message from serverConn is a | |
| -// CommandComplete message with the input tag, and returns an error if it isn't. | |
| -func expectCommandComplete( | |
| - ctx context.Context, serverConn *interceptor.FrontendConn, tag string, | |
| -) error { | |
| - if ctx.Err() != nil { | |
| - return ctx.Err() | |
| - } | |
| - msg, err := serverConn.ReadMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "reading message") | |
| - } | |
| - pgMsg, ok := msg.(*pgproto3.CommandComplete) | |
| - if !ok || string(pgMsg.CommandTag) != tag { | |
| - return errors.Newf("unexpected message: %v", jsonOrRaw(msg)) | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -// expectReadyForQuery expects that the next message from serverConn is a | |
| -// ReadyForQuery message, and returns an error if it isn't. | |
| -func expectReadyForQuery(ctx context.Context, serverConn *interceptor.FrontendConn) error { | |
| - if ctx.Err() != nil { | |
| - return ctx.Err() | |
| - } | |
| - msg, err := serverConn.ReadMsg() | |
| - if err != nil { | |
| - return errors.Wrap(err, "reading message") | |
| - } | |
| - _, ok := msg.(*pgproto3.ReadyForQuery) | |
| - if !ok { | |
| - return errors.Newf("unexpected message: %v", jsonOrRaw(msg)) | |
| - } | |
| - return nil | |
| -} | |
| - | |
| -// jsonOrRaw returns msg in a json string representation if it can be marshaled | |
| -// into one, or in a raw struct string representation otherwise. Only used for | |
| -// displaying better error messages. | |
| -func jsonOrRaw(msg pgproto3.BackendMessage) string { | |
| - m, err := json.Marshal(msg) | |
| - if err != nil { | |
| - return fmt.Sprintf("%v", msg) | |
| - } | |
| - return string(m) | |
| -} | |
| - | |
| -var _ io.Writer = &errWriter{} | |
| - | |
| -// errWriter is an io.Writer that fails whenever a Write call is made. | |
| -type errWriter struct{} | |
| - | |
| -// Write implements the io.Writer interface. | |
| -func (w *errWriter) Write(p []byte) (int, error) { | |
| - return 0, errors.AssertionFailedf("unexpected Write call") | |
| -} | |
| diff --git a/pkg/ccl/sqlproxyccl/conn_migration_test.go b/pkg/ccl/sqlproxyccl/conn_migration_test.go | |
| deleted file mode 100644 | |
| index d8c1d70da4..0000000000 | |
| --- a/pkg/ccl/sqlproxyccl/conn_migration_test.go | |
| +++ /dev/null | |
| @@ -1,831 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Licensed as a CockroachDB Enterprise file under the Cockroach Community | |
| -// License (the "License"); you may not use this file except in compliance with | |
| -// the License. You may obtain a copy of the License at | |
| -// | |
| -// https://github.com/cockroachdb/cockroach/blob/master/licenses/CCL.txt | |
| - | |
| -package sqlproxyccl | |
| - | |
| -import ( | |
| - "bytes" | |
| - "context" | |
| - "io" | |
| - "net" | |
| - "testing" | |
| - "time" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor" | |
| - "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| - "github.com/cockroachdb/errors" | |
| - "github.com/jackc/pgproto3/v2" | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -func TestRunShowTransferState(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - | |
| - t.Run("successful", func(t *testing.T) { | |
| - buf := new(bytes.Buffer) | |
| - err := runShowTransferState(buf, "foo-bar-baz") | |
| - require.NoError(t, err) | |
| - | |
| - backend := pgproto3.NewBackend(pgproto3.NewChunkReader(buf), buf) | |
| - msg, err := backend.Receive() | |
| - require.NoError(t, err) | |
| - m, ok := msg.(*pgproto3.Query) | |
| - require.True(t, ok) | |
| - require.Equal(t, "SHOW TRANSFER STATE WITH 'foo-bar-baz'", m.String) | |
| - }) | |
| - | |
| - t.Run("error", func(t *testing.T) { | |
| - err := runShowTransferState(&errWriter{}, "foo") | |
| - require.Regexp(t, "unexpected Write call", err) | |
| - }) | |
| -} | |
| - | |
| -func TestWaitForShowTransferState(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - t.Run("context_cancelled", func(t *testing.T) { | |
| - tCtx, cancel := context.WithCancel(ctx) | |
| - cancel() | |
| - | |
| - transferErr, state, token, err := waitForShowTransferState(tCtx, nil, nil, "") | |
| - require.True(t, errors.Is(err, context.Canceled)) | |
| - require.Equal(t, "", transferErr) | |
| - require.Equal(t, "", state) | |
| - require.Equal(t, "", token) | |
| - }) | |
| - | |
| - transferStateDataRow := &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - {}, | |
| - []byte("foo-state"), | |
| - []byte("foo-token"), | |
| - []byte("foo-transfer-key"), | |
| - }, | |
| - } | |
| - | |
| - for _, tc := range []struct { | |
| - name string | |
| - sendSequence []pgproto3.BackendMessage | |
| - postValidate func(*testing.T, <-chan pgproto3.BackendMessage) | |
| - err string | |
| - transferErr string | |
| - }{ | |
| - { | |
| - // All irrelevant messages are forwarded to the client. This returns | |
| - // an error when we see ErrorResponse. | |
| - name: "RowDescription/candidate_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - // not RowDescription. | |
| - &pgproto3.BackendKeyData{}, | |
| - // Too large (> 4k). | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("foo1")}, | |
| - {Name: make([]byte, 1<<13 /* 8K */)}, | |
| - }, | |
| - }, | |
| - // Invalid number of columns. | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("foo2")}, | |
| - }, | |
| - }, | |
| - // Invalid column names. | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_foo")}, | |
| - {Name: []byte("session_revival_token_bar")}, | |
| - {Name: []byte("apple")}, | |
| - }, | |
| - }, | |
| - &pgproto3.ErrorResponse{}, | |
| - }, | |
| - err: "ambiguous ErrorResponse", | |
| - postValidate: func(t *testing.T, msgCh <-chan pgproto3.BackendMessage) { | |
| - t.Helper() | |
| - expectMsg(t, msgCh, `"Type":"BackendKeyData"`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription".*"Name":"foo1"`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription".*"Name":"foo2"`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription".*session_state_foo.*session_revival_token_bar`) | |
| - }, | |
| - }, | |
| - { | |
| - name: "DataRow/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - err: `DataRow: unexpected message:.*"Type":"ReadyForQuery"`, | |
| - }, | |
| - { | |
| - name: "DataRow/invalid_response", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - // 3 columns instead of 4. | |
| - Values: [][]byte{ | |
| - {}, | |
| - {}, | |
| - []byte("bar"), | |
| - }, | |
| - }, | |
| - }, | |
| - err: "DataRow: validation failed for message", | |
| - }, | |
| - { | |
| - name: "DataRow/invalid_transfer_key", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - {}, | |
| - {}, | |
| - {}, | |
| - []byte("bar"), | |
| - }, | |
| - }, | |
| - }, | |
| - err: "DataRow: validation failed for message", | |
| - }, | |
| - { | |
| - name: "CommandComplete/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - {}, | |
| - {}, | |
| - {}, | |
| - []byte("foo-transfer-key"), | |
| - }, | |
| - }, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - err: `CommandComplete: unexpected message:.*"Type":"ReadyForQuery"`, | |
| - }, | |
| - { | |
| - name: "CommandComplete/value_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - {}, | |
| - {}, | |
| - {}, | |
| - []byte("foo-transfer-key"), | |
| - }, | |
| - }, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SHOW TRANSFER STATE 2")}, | |
| - }, | |
| - err: `CommandComplete: unexpected message:.*"Type":"CommandComplete".*"CommandTag":"SHOW TRANSFER STATE 2"`, | |
| - }, | |
| - { | |
| - name: "ReadyForQuery/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - {}, | |
| - {}, | |
| - {}, | |
| - []byte("foo-transfer-key"), | |
| - }, | |
| - }, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SHOW TRANSFER STATE 1")}, | |
| - &pgproto3.CommandComplete{}, | |
| - }, | |
| - err: `ReadyForQuery: unexpected message:.*"Type":"CommandComplete"`, | |
| - }, | |
| - { | |
| - // This should be a common case with open transactions. | |
| - name: "transfer_state_error", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{ | |
| - Values: [][]byte{ | |
| - []byte("serialization error"), | |
| - {}, | |
| - {}, | |
| - []byte("foo-transfer-key"), | |
| - }, | |
| - }, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SHOW TRANSFER STATE 1")}, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - transferErr: "serialization error", | |
| - }, | |
| - { | |
| - name: "successful", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.BackendKeyData{}, | |
| - &pgproto3.RowDescription{}, | |
| - &pgproto3.CommandComplete{}, | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("error")}, | |
| - {Name: []byte("session_state_base64")}, | |
| - {Name: []byte("session_revival_token_base64")}, | |
| - {Name: []byte("transfer_key")}, | |
| - }, | |
| - }, | |
| - transferStateDataRow, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SHOW TRANSFER STATE 1")}, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - postValidate: func(t *testing.T, msgCh <-chan pgproto3.BackendMessage) { | |
| - t.Helper() | |
| - expectMsg(t, msgCh, `"Type":"BackendKeyData"`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription"`) | |
| - expectMsg(t, msgCh, `"Type":"CommandComplete"`) | |
| - }, | |
| - }, | |
| - } { | |
| - t.Run(tc.name, func(t *testing.T) { | |
| - serverProxy, server := net.Pipe() | |
| - defer serverProxy.Close() | |
| - defer server.Close() | |
| - | |
| - clientProxy, client := net.Pipe() | |
| - defer clientProxy.Close() | |
| - defer client.Close() | |
| - | |
| - doneCh := make(chan struct{}) | |
| - go func() { | |
| - for _, m := range tc.sendSequence { | |
| - writeServerMsg(server, m) | |
| - } | |
| - close(doneCh) | |
| - }() | |
| - | |
| - msgCh := make(chan pgproto3.BackendMessage, 10) | |
| - go func() { | |
| - fi := interceptor.NewFrontendConn(client) | |
| - for { | |
| - msg, err := fi.ReadMsg() | |
| - if err != nil { | |
| - return | |
| - } | |
| - msgCh <- msg | |
| - } | |
| - }() | |
| - | |
| - transferErr, state, token, err := waitForShowTransferState( | |
| - ctx, | |
| - interceptor.NewFrontendConn(serverProxy), | |
| - clientProxy, | |
| - "foo-transfer-key", | |
| - ) | |
| - if tc.err == "" { | |
| - require.NoError(t, err) | |
| - if tc.transferErr != "" { | |
| - require.Equal(t, tc.transferErr, transferErr) | |
| - } else { | |
| - require.Equal(t, "", transferErr) | |
| - require.Equal(t, "foo-state", state) | |
| - require.Equal(t, "foo-token", token) | |
| - | |
| - // Ensure that returned strings are copied. Alternatively, | |
| - // we could also check pointers using encoding.UnsafeConvertStringToBytes. | |
| - transferStateDataRow.Values[0] = []byte("x") | |
| - transferStateDataRow.Values[1][1] = '-' | |
| - transferStateDataRow.Values[2][1] = '-' | |
| - require.Equal(t, "", transferErr) | |
| - require.Equal(t, "foo-state", state) | |
| - require.Equal(t, "foo-token", token) | |
| - } | |
| - require.Eventually(t, func() bool { | |
| - select { | |
| - case <-doneCh: | |
| - return true | |
| - default: | |
| - return false | |
| - } | |
| - }, 5*time.Second, 100*time.Millisecond, "require doneCh to be closed") | |
| - } else { | |
| - require.Regexp(t, tc.err, err) | |
| - } | |
| - | |
| - // Verify that forwarding was correct. | |
| - if tc.postValidate != nil { | |
| - tc.postValidate(t, msgCh) | |
| - } | |
| - }) | |
| - } | |
| -} | |
| - | |
| -func TestRunAndWaitForDeserializeSession(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - t.Run("write_failed", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - r.Close() | |
| - w.Close() | |
| - err := runAndWaitForDeserializeSession(ctx, | |
| - interceptor.NewFrontendConn(r), "foo") | |
| - require.Regexp(t, "closed pipe", err) | |
| - }) | |
| - | |
| - for _, tc := range []struct { | |
| - name string | |
| - sendSequence []pgproto3.BackendMessage | |
| - err string | |
| - }{ | |
| - { | |
| - name: "RowDescription/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.BackendKeyData{}, | |
| - }, | |
| - err: "RowDescription: forwarding message: unexpected Write call", | |
| - }, | |
| - { | |
| - name: "RowDescription/column_mismatch/length", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{}, | |
| - }, | |
| - err: "RowDescription: writing message: unexpected Write call", | |
| - }, | |
| - { | |
| - name: "RowDescription/column_mismatch/name", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{{Name: []byte("bar")}}}, | |
| - }, | |
| - err: "RowDescription: writing message: unexpected Write call", | |
| - }, | |
| - { | |
| - name: "DataRow/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - err: `DataRow: unexpected message:.*"Type":"ReadyForQuery"`, | |
| - }, | |
| - { | |
| - name: "DataRow/column_mismatch/length", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{}, | |
| - }, | |
| - err: `DataRow: validation failed for message`, | |
| - }, | |
| - { | |
| - name: "DataRow/column_mismatch/value", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{Values: [][]byte{[]byte("temp")}}, | |
| - }, | |
| - err: "DataRow: validation failed for message", | |
| - }, | |
| - { | |
| - name: "CommandComplete/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{Values: [][]byte{[]byte("t")}}, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - err: `CommandComplete: unexpected message:.*"Type":"ReadyForQuery"`, | |
| - }, | |
| - { | |
| - name: "CommandComplete/value_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{Values: [][]byte{[]byte("t")}}, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SELECT 2")}, | |
| - }, | |
| - err: `CommandComplete: unexpected message:.*"Type":"CommandComplete".*"CommandTag":"SELECT 2"`, | |
| - }, | |
| - { | |
| - name: "ReadyForQuery/type_mismatch", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{Values: [][]byte{[]byte("t")}}, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SELECT 1")}, | |
| - &pgproto3.CommandComplete{}, | |
| - }, | |
| - err: `ReadyForQuery: unexpected message:.*"Type":"CommandComplete"`, | |
| - }, | |
| - { | |
| - name: "successful", | |
| - sendSequence: []pgproto3.BackendMessage{ | |
| - &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("crdb_internal.deserialize_session")}, | |
| - }, | |
| - }, | |
| - &pgproto3.DataRow{Values: [][]byte{[]byte("t")}}, | |
| - &pgproto3.CommandComplete{CommandTag: []byte("SELECT 1")}, | |
| - &pgproto3.ReadyForQuery{}, | |
| - }, | |
| - err: "", | |
| - }, | |
| - } { | |
| - t.Run(tc.name, func(t *testing.T) { | |
| - serverProxy, server := net.Pipe() | |
| - defer serverProxy.Close() | |
| - defer server.Close() | |
| - doneCh := make(chan struct{}) | |
| - go func() { | |
| - for _, m := range tc.sendSequence { | |
| - writeServerMsg(server, m) | |
| - } | |
| - close(doneCh) | |
| - }() | |
| - | |
| - msgCh := make(chan pgproto3.FrontendMessage, 1) | |
| - go func() { | |
| - backend := interceptor.NewBackendConn(server) | |
| - msg, _ := backend.ReadMsg() | |
| - msgCh <- msg | |
| - }() | |
| - | |
| - err := runAndWaitForDeserializeSession(ctx, | |
| - interceptor.NewFrontendConn(serverProxy), "foo-transfer-key") | |
| - if tc.err == "" { | |
| - require.NoError(t, err) | |
| - } else { | |
| - require.Regexp(t, tc.err, err) | |
| - } | |
| - | |
| - require.Eventually(t, func() bool { | |
| - select { | |
| - case <-doneCh: | |
| - return true | |
| - default: | |
| - return false | |
| - } | |
| - }, 5*time.Second, 100*time.Millisecond, "require doneCh to be closed") | |
| - | |
| - msg := <-msgCh | |
| - m, ok := msg.(*pgproto3.Query) | |
| - require.True(t, ok) | |
| - const queryStr = "SELECT crdb_internal.deserialize_session(decode('foo-transfer-key', 'base64'))" | |
| - require.Equal(t, queryStr, m.String) | |
| - }) | |
| - } | |
| -} | |
| - | |
| -func TestWaitForSmallRowDescription(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - t.Run("context_cancelled", func(t *testing.T) { | |
| - tCtx, cancel := context.WithCancel(ctx) | |
| - cancel() | |
| - | |
| - err := waitForSmallRowDescription(tCtx, nil, nil, nil) | |
| - require.EqualError(t, err, context.Canceled.Error()) | |
| - }) | |
| - | |
| - t.Run("peek_error", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - r.Close() | |
| - w.Close() | |
| - | |
| - err := waitForSmallRowDescription(ctx, interceptor.NewFrontendConn(r), nil, nil) | |
| - require.Regexp(t, "peeking message", err) | |
| - }) | |
| - | |
| - t.Run("ambiguous_error", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.ErrorResponse{}) | |
| - }() | |
| - | |
| - err := waitForSmallRowDescription(ctx, interceptor.NewFrontendConn(r), nil, nil) | |
| - require.Regexp(t, "ambiguous ErrorResponse.*ErrorResponse", err) | |
| - }) | |
| - | |
| - t.Run("successful", func(t *testing.T) { | |
| - serverProxy, server := net.Pipe() | |
| - defer serverProxy.Close() | |
| - defer server.Close() | |
| - | |
| - clientProxy, client := net.Pipe() | |
| - defer clientProxy.Close() | |
| - defer client.Close() | |
| - | |
| - go func() { | |
| - // Not RowDescription. | |
| - writeServerMsg(server, &pgproto3.BackendKeyData{ProcessID: 42}) | |
| - // Too large (> 4k bytes). | |
| - writeServerMsg(server, &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("foo1")}, | |
| - {Name: make([]byte, 1<<13 /* 8K */)}, | |
| - }, | |
| - }) | |
| - // Mismatch. | |
| - writeServerMsg(server, &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("foo2")}, | |
| - {Name: []byte("foo3")}, | |
| - }, | |
| - }) | |
| - // Match. | |
| - writeServerMsg(server, &pgproto3.RowDescription{ | |
| - Fields: []pgproto3.FieldDescription{ | |
| - {Name: []byte("foo1")}, | |
| - }, | |
| - }) | |
| - }() | |
| - | |
| - msgCh := make(chan pgproto3.BackendMessage, 10) | |
| - go func() { | |
| - fi := interceptor.NewFrontendConn(client) | |
| - for { | |
| - msg, err := fi.ReadMsg() | |
| - if err != nil { | |
| - return | |
| - } | |
| - msgCh <- msg | |
| - } | |
| - }() | |
| - | |
| - err := waitForSmallRowDescription( | |
| - ctx, | |
| - interceptor.NewFrontendConn(serverProxy), | |
| - clientProxy, | |
| - func(m *pgproto3.RowDescription) bool { | |
| - return len(m.Fields) == 1 && string(m.Fields[0].Name) == "foo1" | |
| - }, | |
| - ) | |
| - require.Nil(t, err) | |
| - | |
| - // Verify that forwarding was correct. | |
| - expectMsg(t, msgCh, `"Type":"BackendKeyData".*"ProcessID":42`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription".*"Name":"foo1"`) | |
| - expectMsg(t, msgCh, `"Type":"RowDescription".*"Name":"foo2".*"Name":"foo3"`) | |
| - }) | |
| -} | |
| - | |
| -func TestExpectDataRow(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - falseValidateFn := func(m *pgproto3.DataRow) bool { return false } | |
| - | |
| - t.Run("context_cancelled", func(t *testing.T) { | |
| - tCtx, cancel := context.WithCancel(ctx) | |
| - cancel() | |
| - | |
| - err := expectDataRow(tCtx, nil, falseValidateFn) | |
| - require.EqualError(t, err, context.Canceled.Error()) | |
| - }) | |
| - | |
| - t.Run("read_error", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - r.Close() | |
| - w.Close() | |
| - | |
| - err := expectDataRow(ctx, interceptor.NewFrontendConn(r), falseValidateFn) | |
| - require.Regexp(t, "reading message", err) | |
| - }) | |
| - | |
| - t.Run("type_mismatch", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.ReadyForQuery{}) | |
| - }() | |
| - | |
| - err := expectDataRow(ctx, interceptor.NewFrontendConn(r), falseValidateFn) | |
| - require.Regexp(t, "unexpected message.*ReadyForQuery", err) | |
| - }) | |
| - | |
| - t.Run("validation_failed", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.DataRow{}) | |
| - }() | |
| - | |
| - err := expectDataRow(ctx, interceptor.NewFrontendConn(r), falseValidateFn) | |
| - require.Regexp(t, "validation failed for message.*DataRow", err) | |
| - }) | |
| - | |
| - t.Run("successful", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.DataRow{Values: [][]byte{[]byte("foo")}}) | |
| - }() | |
| - | |
| - err := expectDataRow( | |
| - ctx, | |
| - interceptor.NewFrontendConn(r), | |
| - func(m *pgproto3.DataRow) bool { | |
| - return len(m.Values) == 1 && string(m.Values[0]) == "foo" | |
| - }, | |
| - ) | |
| - require.Nil(t, err) | |
| - }) | |
| -} | |
| - | |
| -func TestExpectCommandComplete(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - t.Run("context_cancelled", func(t *testing.T) { | |
| - tCtx, cancel := context.WithCancel(ctx) | |
| - cancel() | |
| - | |
| - err := expectCommandComplete(tCtx, nil, "") | |
| - require.EqualError(t, err, context.Canceled.Error()) | |
| - }) | |
| - | |
| - t.Run("read_error", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - r.Close() | |
| - w.Close() | |
| - | |
| - err := expectCommandComplete(ctx, interceptor.NewFrontendConn(r), "") | |
| - require.Regexp(t, "reading message", err) | |
| - }) | |
| - | |
| - t.Run("type_mismatch", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.ReadyForQuery{}) | |
| - }() | |
| - | |
| - err := expectCommandComplete(ctx, interceptor.NewFrontendConn(r), "") | |
| - require.Regexp(t, "unexpected message.*ReadyForQuery", err) | |
| - }) | |
| - | |
| - t.Run("tag_mismatch", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.CommandComplete{CommandTag: []byte("foo")}) | |
| - }() | |
| - | |
| - err := expectCommandComplete(ctx, interceptor.NewFrontendConn(r), "bar") | |
| - require.Regexp(t, "unexpected message.*CommandComplete.*CommandTag.*foo", err) | |
| - }) | |
| - | |
| - t.Run("successful", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.CommandComplete{CommandTag: []byte("SELECT 1")}) | |
| - }() | |
| - | |
| - err := expectCommandComplete(ctx, interceptor.NewFrontendConn(r), "SELECT 1") | |
| - require.Nil(t, err) | |
| - }) | |
| -} | |
| - | |
| -func TestExpectReadyForQuery(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - | |
| - t.Run("context_cancelled", func(t *testing.T) { | |
| - tCtx, cancel := context.WithCancel(ctx) | |
| - cancel() | |
| - | |
| - err := expectReadyForQuery(tCtx, nil) | |
| - require.EqualError(t, err, context.Canceled.Error()) | |
| - }) | |
| - | |
| - t.Run("read_error", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - r.Close() | |
| - w.Close() | |
| - | |
| - err := expectReadyForQuery(ctx, interceptor.NewFrontendConn(r)) | |
| - require.Regexp(t, "reading message", err) | |
| - }) | |
| - | |
| - t.Run("type_mismatch", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.ErrorResponse{}) | |
| - }() | |
| - | |
| - err := expectReadyForQuery(ctx, interceptor.NewFrontendConn(r)) | |
| - require.Regexp(t, "unexpected message.*ErrorResponse", err) | |
| - }) | |
| - | |
| - t.Run("successful", func(t *testing.T) { | |
| - r, w := net.Pipe() | |
| - defer r.Close() | |
| - defer w.Close() | |
| - | |
| - go func() { | |
| - writeServerMsg(w, &pgproto3.ReadyForQuery{TxStatus: 'I'}) | |
| - }() | |
| - | |
| - err := expectReadyForQuery(ctx, interceptor.NewFrontendConn(r)) | |
| - require.Nil(t, err) | |
| - }) | |
| -} | |
| - | |
| -func writeServerMsg(w io.Writer, msg pgproto3.BackendMessage) { | |
| - _, _ = w.Write(msg.Encode(nil)) | |
| -} | |
| - | |
| -func expectMsg(t *testing.T, msgCh <-chan pgproto3.BackendMessage, match string) { | |
| - t.Helper() | |
| - msg := <-msgCh | |
| - require.Regexp(t, match, jsonOrRaw(msg)) | |
| -} | |
| diff --git a/pkg/ccl/sqlproxyccl/forwarder.go b/pkg/ccl/sqlproxyccl/forwarder.go | |
| index 67040cef88..9810ef695a 100644 | |
| --- a/pkg/ccl/sqlproxyccl/forwarder.go | |
| +++ b/pkg/ccl/sqlproxyccl/forwarder.go | |
| @@ -31,23 +31,31 @@ type forwarder struct { | |
| ctx context.Context | |
| ctxCancel context.CancelFunc | |
| - // clientConn and serverConn provide a convenient way to read and forward | |
| - // Postgres messages, while minimizing IO reads and memory allocations. | |
| - // | |
| - // clientConn is set once during initialization, and stays the same | |
| - // throughout the lifetime of the forwarder. | |
| - // | |
| // serverConn is only set after the authentication phase for the initial | |
| // connection. In the context of a connection migration, serverConn is only | |
| - // replaced once the session has successfully been deserialized, and the old | |
| - // connection will be closed. Whenever serverConn gets updated, both | |
| - // clientMessageTypeSent and isServerMsgReadyReceived fields have to reset | |
| - // to their initial values. | |
| + // replaced once the session has successfully been deserialized, and the | |
| + // old connection will be closed. | |
| // | |
| // All reads from these connections must go through the interceptors. It is | |
| - // not safe to call Read directly as the interceptors may have buffered data. | |
| - clientConn *interceptor.BackendConn // client <-> proxy | |
| - serverConn *interceptor.FrontendConn // proxy <-> server | |
| + // not safe to read from these directly as the interceptors may have | |
| + // buffered data. | |
| + clientConn net.Conn // client <-> proxy | |
| + serverConn net.Conn // proxy <-> server | |
| + | |
| + // clientInterceptor and serverInterceptor provides a convenient way to | |
| + // read and forward Postgres messages, while minimizing IO reads and memory | |
| + // allocations. | |
| + // | |
| + // These interceptors have to match clientConn and serverConn. See comment | |
| + // above on when those fields will be updated. | |
| + // | |
| + // TODO(jaylim-crl): Add updater functions that sets both conn and | |
| + // interceptor fields at the same time. At the moment, there's no use case | |
| + // besides the forward function. When connection migration happens, we | |
| + // will need to create a new serverInterceptor. We should remember to close | |
| + // old serverConn as well. | |
| + clientInterceptor *interceptor.BackendInterceptor // clientConn -> serverConn | |
| + serverInterceptor *interceptor.FrontendInterceptor // serverConn -> clientConn | |
| // errChan is a buffered channel that contains the first forwarder error. | |
| // This channel may receive nil errors. | |
| @@ -82,8 +90,9 @@ func forward(ctx context.Context, clientConn, serverConn net.Conn) *forwarder { | |
| // TODO(jaylim-crl): Check for transfer state here. | |
| return nil | |
| }) | |
| - f.clientConn = interceptor.NewBackendConn(clientConn) | |
| - f.serverConn = interceptor.NewFrontendConn(serverConn) | |
| + | |
| + f.setClientConn(clientConn) | |
| + f.setServerConn(serverConn) | |
| // Start request (client to server) and response (server to client) | |
| // processors. We will copy all pgwire messages/ from client to server | |
| @@ -128,7 +137,7 @@ func (f *forwarder) Close() { | |
| // gets triggered when context is cancelled. | |
| func (f *forwarder) handleClientToServer() error { | |
| for f.ctx.Err() == nil { | |
| - if _, err := f.clientConn.ForwardMsg(f.serverConn); err != nil { | |
| + if _, err := f.clientInterceptor.ForwardMsg(f.serverConn); err != nil { | |
| return err | |
| } | |
| } | |
| @@ -141,13 +150,31 @@ func (f *forwarder) handleClientToServer() error { | |
| // Read, we will unblock that by closing serverConn through f.Close(). | |
| func (f *forwarder) handleServerToClient() error { | |
| for f.ctx.Err() == nil { | |
| - if _, err := f.serverConn.ForwardMsg(f.clientConn); err != nil { | |
| + if _, err := f.serverInterceptor.ForwardMsg(f.clientConn); err != nil { | |
| return err | |
| } | |
| } | |
| return f.ctx.Err() | |
| } | |
| +// setClientConn is a convenient helper to update clientConn, and will also | |
| +// create a matching interceptor for the given connection. It is the caller's | |
| +// responsibility to close the old connection before calling this, or there | |
| +// may be a leak. | |
| +func (f *forwarder) setClientConn(clientConn net.Conn) { | |
| + f.clientConn = clientConn | |
| + f.clientInterceptor = interceptor.NewBackendInterceptor(f.clientConn) | |
| +} | |
| + | |
| +// setServerConn is a convenient helper to update serverConn, and will also | |
| +// create a matching interceptor for the given connection. It is the caller's | |
| +// responsibility to close the old connection before calling this, or there | |
| +// may be a leak. | |
| +func (f *forwarder) setServerConn(serverConn net.Conn) { | |
| + f.serverConn = serverConn | |
| + f.serverInterceptor = interceptor.NewFrontendInterceptor(f.serverConn) | |
| +} | |
| + | |
| // wrapClientToServerError overrides client to server errors for external | |
| // consumption. | |
| // | |
| diff --git a/pkg/ccl/sqlproxyccl/forwarder_test.go b/pkg/ccl/sqlproxyccl/forwarder_test.go | |
| index 0b398d75d1..0acdf12e5c 100644 | |
| --- a/pkg/ccl/sqlproxyccl/forwarder_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/forwarder_test.go | |
| @@ -9,6 +9,7 @@ | |
| package sqlproxyccl | |
| import ( | |
| + "bytes" | |
| "context" | |
| "net" | |
| "testing" | |
| @@ -187,6 +188,62 @@ func TestForwarder_Close(t *testing.T) { | |
| require.EqualError(t, f.ctx.Err(), context.Canceled.Error()) | |
| } | |
| +func TestForwarder_setClientConn(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + f := &forwarder{serverConn: nil, serverInterceptor: nil} | |
| + | |
| + w, r := net.Pipe() | |
| + defer w.Close() | |
| + defer r.Close() | |
| + | |
| + f.setClientConn(r) | |
| + require.Equal(t, r, f.clientConn) | |
| + | |
| + dst := new(bytes.Buffer) | |
| + errChan := make(chan error, 1) | |
| + go func() { | |
| + _, err := f.clientInterceptor.ForwardMsg(dst) | |
| + errChan <- err | |
| + }() | |
| + | |
| + _, err := w.Write((&pgproto3.Query{String: "SELECT 1"}).Encode(nil)) | |
| + require.NoError(t, err) | |
| + | |
| + // Block until message has been forwarded. This checks that we are creating | |
| + // our interceptor properly. | |
| + err = <-errChan | |
| + require.NoError(t, err) | |
| + require.Equal(t, 14, dst.Len()) | |
| +} | |
| + | |
| +func TestForwarder_setServerConn(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + f := &forwarder{serverConn: nil, serverInterceptor: nil} | |
| + | |
| + w, r := net.Pipe() | |
| + defer w.Close() | |
| + defer r.Close() | |
| + | |
| + f.setServerConn(r) | |
| + require.Equal(t, r, f.serverConn) | |
| + | |
| + dst := new(bytes.Buffer) | |
| + errChan := make(chan error, 1) | |
| + go func() { | |
| + _, err := f.serverInterceptor.ForwardMsg(dst) | |
| + errChan <- err | |
| + }() | |
| + | |
| + _, err := w.Write((&pgproto3.ReadyForQuery{TxStatus: 'I'}).Encode(nil)) | |
| + require.NoError(t, err) | |
| + | |
| + // Block until message has been forwarded. This checks that we are creating | |
| + // our interceptor properly. | |
| + err = <-errChan | |
| + require.NoError(t, err) | |
| + require.Equal(t, 6, dst.Len()) | |
| +} | |
| + | |
| func TestWrapClientToServerError(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/BUILD.bazel b/pkg/ccl/sqlproxyccl/interceptor/BUILD.bazel | |
| index 2052d9670a..11771fdc56 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/BUILD.bazel | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/BUILD.bazel | |
| @@ -3,10 +3,10 @@ load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test") | |
| go_library( | |
| name = "interceptor", | |
| srcs = [ | |
| - "backend_conn.go", | |
| + "backend_interceptor.go", | |
| "base.go", | |
| "chunkreader.go", | |
| - "frontend_conn.go", | |
| + "frontend_interceptor.go", | |
| ], | |
| importpath = "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor", | |
| visibility = ["//visibility:public"], | |
| @@ -21,10 +21,10 @@ go_library( | |
| go_test( | |
| name = "interceptor_test", | |
| srcs = [ | |
| - "backend_conn_test.go", | |
| + "backend_interceptor_test.go", | |
| "base_test.go", | |
| "chunkreader_test.go", | |
| - "frontend_conn_test.go", | |
| + "frontend_interceptor_test.go", | |
| "interceptor_test.go", | |
| ], | |
| embed = [":interceptor"], | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/backend_conn.go b/pkg/ccl/sqlproxyccl/interceptor/backend_interceptor.go | |
| similarity index 60% | |
| rename from pkg/ccl/sqlproxyccl/interceptor/backend_conn.go | |
| rename to pkg/ccl/sqlproxyccl/interceptor/backend_interceptor.go | |
| index 6f6977786a..0611b267a4 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/backend_conn.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/backend_interceptor.go | |
| @@ -10,33 +10,26 @@ package interceptor | |
| import ( | |
| "io" | |
| - "net" | |
| "github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgwirebase" | |
| "github.com/jackc/pgproto3/v2" | |
| ) | |
| -// BackendConn is a server interceptor for the Postgres backend protocol. | |
| -// This will be used for the connection between client and proxy. | |
| -type BackendConn struct { | |
| - net.Conn | |
| - interceptor *pgInterceptor | |
| -} | |
| +// BackendInterceptor is a server int/erceptor for the Postgres backend protocol. | |
| +type BackendInterceptor pgInterceptor | |
| -// NewBackendConn creates a BackendConn using the default buffer size of 8KB. | |
| -func NewBackendConn(conn net.Conn) *BackendConn { | |
| - return &BackendConn{ | |
| - Conn: conn, | |
| - interceptor: newPgInterceptor(conn, defaultBufferSize), | |
| - } | |
| +// NewBackendInterceptor creates a BackendInterceptor using the default buffer | |
| +// size of 8K bytes. | |
| +func NewBackendInterceptor(src io.Reader) *BackendInterceptor { | |
| + return (*BackendInterceptor)(newPgInterceptor(src, defaultBufferSize)) | |
| } | |
| // PeekMsg returns the header of the current pgwire message without advancing | |
| // the interceptor. | |
| // | |
| // See pgInterceptor.PeekMsg for more information. | |
| -func (c *BackendConn) PeekMsg() (typ pgwirebase.ClientMessageType, size int, err error) { | |
| - byteType, size, err := c.interceptor.PeekMsg() | |
| +func (bi *BackendInterceptor) PeekMsg() (typ pgwirebase.ClientMessageType, size int, err error) { | |
| + byteType, size, err := (*pgInterceptor)(bi).PeekMsg() | |
| return pgwirebase.ClientMessageType(byteType), size, err | |
| } | |
| @@ -44,8 +37,8 @@ func (c *BackendConn) PeekMsg() (typ pgwirebase.ClientMessageType, size int, err | |
| // This also advances the interceptor to the next message. | |
| // | |
| // See pgInterceptor.ReadMsg for more information. | |
| -func (c *BackendConn) ReadMsg() (msg pgproto3.FrontendMessage, err error) { | |
| - msgBytes, err := c.interceptor.ReadMsg() | |
| +func (bi *BackendInterceptor) ReadMsg() (msg pgproto3.FrontendMessage, err error) { | |
| + msgBytes, err := (*pgInterceptor)(bi).ReadMsg() | |
| if err != nil { | |
| return nil, err | |
| } | |
| @@ -57,6 +50,6 @@ func (c *BackendConn) ReadMsg() (msg pgproto3.FrontendMessage, err error) { | |
| // decoding, and advances the interceptor to the next message. | |
| // | |
| // See pgInterceptor.ForwardMsg for more information. | |
| -func (c *BackendConn) ForwardMsg(dst io.Writer) (n int, err error) { | |
| - return c.interceptor.ForwardMsg(dst) | |
| +func (bi *BackendInterceptor) ForwardMsg(dst io.Writer) (n int, err error) { | |
| + return (*pgInterceptor)(bi).ForwardMsg(dst) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/backend_conn_test.go b/pkg/ccl/sqlproxyccl/interceptor/backend_interceptor_test.go | |
| similarity index 66% | |
| rename from pkg/ccl/sqlproxyccl/interceptor/backend_conn_test.go | |
| rename to pkg/ccl/sqlproxyccl/interceptor/backend_interceptor_test.go | |
| index 2b4de17e8f..50fba567d0 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/backend_conn_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/backend_interceptor_test.go | |
| @@ -10,8 +10,6 @@ package interceptor_test | |
| import ( | |
| "bytes" | |
| - "io" | |
| - "net" | |
| "testing" | |
| "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor" | |
| @@ -21,44 +19,38 @@ import ( | |
| "github.com/stretchr/testify/require" | |
| ) | |
| -// TestBackendConn tests the BackendConn. Note that the tests here are shallow. | |
| -// For detailed ones, see the tests for the internal interceptor in base_test.go. | |
| -func TestBackendConn(t *testing.T) { | |
| +// TestBackendInterceptor tests the BackendInterceptor. Note that the tests | |
| +// here are shallow. For detailed ones, see the tests for the internal | |
| +// interceptor in base_test.go. | |
| +func TestBackendInterceptor(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| q := (&pgproto3.Query{String: "SELECT 1"}).Encode(nil) | |
| - writeAsync := func(t *testing.T, w io.Writer) <-chan error { | |
| + buildSrc := func(t *testing.T, count int) *bytes.Buffer { | |
| t.Helper() | |
| - errCh := make(chan error, 1) | |
| - go func() { | |
| - _, err := w.Write(q) | |
| - errCh <- err | |
| - }() | |
| - return errCh | |
| + src := new(bytes.Buffer) | |
| + _, err := src.Write(q) | |
| + require.NoError(t, err) | |
| + return src | |
| } | |
| t.Run("PeekMsg returns the right message type", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| - bi := interceptor.NewBackendConn(r) | |
| + bi := interceptor.NewBackendInterceptor(src) | |
| require.NotNil(t, bi) | |
| typ, size, err := bi.PeekMsg() | |
| require.NoError(t, err) | |
| require.Equal(t, pgwirebase.ClientMsgSimpleQuery, typ) | |
| require.Equal(t, 14, size) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| t.Run("ReadMsg decodes the message correctly", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| - bi := interceptor.NewBackendConn(r) | |
| + bi := interceptor.NewBackendInterceptor(src) | |
| require.NotNil(t, bi) | |
| msg, err := bi.ReadMsg() | |
| @@ -66,25 +58,18 @@ func TestBackendConn(t *testing.T) { | |
| rmsg, ok := msg.(*pgproto3.Query) | |
| require.True(t, ok) | |
| require.Equal(t, "SELECT 1", rmsg.String) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| t.Run("ForwardMsg forwards data to dst", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| dst := new(bytes.Buffer) | |
| - bi := interceptor.NewBackendConn(r) | |
| + bi := interceptor.NewBackendInterceptor(src) | |
| require.NotNil(t, bi) | |
| n, err := bi.ForwardMsg(dst) | |
| require.NoError(t, err) | |
| require.Equal(t, 14, n) | |
| require.Equal(t, 14, dst.Len()) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/base.go b/pkg/ccl/sqlproxyccl/interceptor/base.go | |
| index b9c113600f..c59945cb0b 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/base.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/base.go | |
| @@ -26,7 +26,7 @@ const pgHeaderSizeBytes = 5 | |
| // chosen to match Postgres' send and receive buffer sizes. | |
| // | |
| // See: https://github.com/postgres/postgres/blob/249d64999615802752940e017ee5166e726bc7cd/src/backend/libpq/pqcomm.c#L134-L135. | |
| -const defaultBufferSize = 1 << 13 // 8K | |
| +const defaultBufferSize = 2 << 13 // 8K | |
| // ErrProtocolError indicates that the packets are malformed, and are not as | |
| // expected. | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/base_test.go b/pkg/ccl/sqlproxyccl/interceptor/base_test.go | |
| index 28463e9af9..aca49e46f9 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/base_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/base_test.go | |
| @@ -33,8 +33,8 @@ func TestNewPgInterceptor(t *testing.T) { | |
| bufSize int | |
| normalizedBufSize int | |
| }{ | |
| - {-1, 8192}, | |
| - {pgHeaderSizeBytes - 1, 8192}, | |
| + {-1, defaultBufferSize}, | |
| + {pgHeaderSizeBytes - 1, defaultBufferSize}, | |
| {pgHeaderSizeBytes, pgHeaderSizeBytes}, | |
| {1024, 1024}, | |
| } { | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/frontend_conn.go b/pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor.go | |
| similarity index 60% | |
| rename from pkg/ccl/sqlproxyccl/interceptor/frontend_conn.go | |
| rename to pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor.go | |
| index 6967a1ab78..cd4a5ca9ca 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/frontend_conn.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor.go | |
| @@ -10,33 +10,26 @@ package interceptor | |
| import ( | |
| "io" | |
| - "net" | |
| "github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgwirebase" | |
| "github.com/jackc/pgproto3/v2" | |
| ) | |
| -// FrontendConn is a client interceptor for the Postgres frontend protocol. | |
| -// This will be used for the connection between proxy and server. | |
| -type FrontendConn struct { | |
| - net.Conn | |
| - interceptor *pgInterceptor | |
| -} | |
| +// FrontendInterceptor is a client interceptor for the Postgres frontend protocol. | |
| +type FrontendInterceptor pgInterceptor | |
| -// NewFrontendConn creates a FrontendConn using the default buffer size of 8KB. | |
| -func NewFrontendConn(conn net.Conn) *FrontendConn { | |
| - return &FrontendConn{ | |
| - Conn: conn, | |
| - interceptor: newPgInterceptor(conn, defaultBufferSize), | |
| - } | |
| +// NewFrontendInterceptor creates a FrontendInterceptor using the default buffer | |
| +// size of 8K bytes. | |
| +func NewFrontendInterceptor(src io.Reader) *FrontendInterceptor { | |
| + return (*FrontendInterceptor)(newPgInterceptor(src, defaultBufferSize)) | |
| } | |
| // PeekMsg returns the header of the current pgwire message without advancing | |
| // the interceptor. | |
| // | |
| // See pgInterceptor.PeekMsg for more information. | |
| -func (c *FrontendConn) PeekMsg() (typ pgwirebase.ServerMessageType, size int, err error) { | |
| - byteType, size, err := c.interceptor.PeekMsg() | |
| +func (fi *FrontendInterceptor) PeekMsg() (typ pgwirebase.ServerMessageType, size int, err error) { | |
| + byteType, size, err := (*pgInterceptor)(fi).PeekMsg() | |
| return pgwirebase.ServerMessageType(byteType), size, err | |
| } | |
| @@ -44,8 +37,8 @@ func (c *FrontendConn) PeekMsg() (typ pgwirebase.ServerMessageType, size int, er | |
| // This also advances the interceptor to the next message. | |
| // | |
| // See pgInterceptor.ReadMsg for more information. | |
| -func (c *FrontendConn) ReadMsg() (msg pgproto3.BackendMessage, err error) { | |
| - msgBytes, err := c.interceptor.ReadMsg() | |
| +func (fi *FrontendInterceptor) ReadMsg() (msg pgproto3.BackendMessage, err error) { | |
| + msgBytes, err := (*pgInterceptor)(fi).ReadMsg() | |
| if err != nil { | |
| return nil, err | |
| } | |
| @@ -57,6 +50,6 @@ func (c *FrontendConn) ReadMsg() (msg pgproto3.BackendMessage, err error) { | |
| // decoding, and advances the interceptor to the next message. | |
| // | |
| // See pgInterceptor.ForwardMsg for more information. | |
| -func (c *FrontendConn) ForwardMsg(dst io.Writer) (n int, err error) { | |
| - return c.interceptor.ForwardMsg(dst) | |
| +func (fi *FrontendInterceptor) ForwardMsg(dst io.Writer) (n int, err error) { | |
| + return (*pgInterceptor)(fi).ForwardMsg(dst) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/frontend_conn_test.go b/pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor_test.go | |
| similarity index 66% | |
| rename from pkg/ccl/sqlproxyccl/interceptor/frontend_conn_test.go | |
| rename to pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor_test.go | |
| index 730a4d2860..7e1b03f50d 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/frontend_conn_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/frontend_interceptor_test.go | |
| @@ -10,8 +10,6 @@ package interceptor_test | |
| import ( | |
| "bytes" | |
| - "io" | |
| - "net" | |
| "testing" | |
| "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor" | |
| @@ -21,44 +19,38 @@ import ( | |
| "github.com/stretchr/testify/require" | |
| ) | |
| -// TestFrontendConn tests the FrontendConn. Note that the tests here are shallow. | |
| -// For detailed ones, see the tests for the internal interceptor in base_test.go. | |
| -func TestFrontendConn(t *testing.T) { | |
| +// TestFrontendInterceptor tests the FrontendInterceptor. Note that the tests | |
| +// here are shallow. For detailed ones, see the tests for the internal | |
| +// interceptor in base_test.go. | |
| +func TestFrontendInterceptor(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| q := (&pgproto3.ReadyForQuery{TxStatus: 'I'}).Encode(nil) | |
| - writeAsync := func(t *testing.T, w io.Writer) <-chan error { | |
| + buildSrc := func(t *testing.T, count int) *bytes.Buffer { | |
| t.Helper() | |
| - errCh := make(chan error, 1) | |
| - go func() { | |
| - _, err := w.Write(q) | |
| - errCh <- err | |
| - }() | |
| - return errCh | |
| + src := new(bytes.Buffer) | |
| + _, err := src.Write(q) | |
| + require.NoError(t, err) | |
| + return src | |
| } | |
| t.Run("PeekMsg returns the right message type", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| - fi := interceptor.NewFrontendConn(r) | |
| + fi := interceptor.NewFrontendInterceptor(src) | |
| require.NotNil(t, fi) | |
| typ, size, err := fi.PeekMsg() | |
| require.NoError(t, err) | |
| require.Equal(t, pgwirebase.ServerMsgReady, typ) | |
| require.Equal(t, 6, size) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| t.Run("ReadMsg decodes the message correctly", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| - fi := interceptor.NewFrontendConn(r) | |
| + fi := interceptor.NewFrontendInterceptor(src) | |
| require.NotNil(t, fi) | |
| msg, err := fi.ReadMsg() | |
| @@ -66,25 +58,18 @@ func TestFrontendConn(t *testing.T) { | |
| rmsg, ok := msg.(*pgproto3.ReadyForQuery) | |
| require.True(t, ok) | |
| require.Equal(t, byte('I'), rmsg.TxStatus) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| t.Run("ForwardMsg forwards data to dst", func(t *testing.T) { | |
| - w, r := net.Pipe() | |
| - errCh := writeAsync(t, w) | |
| + src := buildSrc(t, 1) | |
| dst := new(bytes.Buffer) | |
| - fi := interceptor.NewFrontendConn(r) | |
| + fi := interceptor.NewFrontendInterceptor(src) | |
| require.NotNil(t, fi) | |
| n, err := fi.ForwardMsg(dst) | |
| require.NoError(t, err) | |
| require.Equal(t, 6, n) | |
| require.Equal(t, 6, dst.Len()) | |
| - | |
| - err = <-errCh | |
| - require.Nil(t, err) | |
| }) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/interceptor/interceptor_test.go b/pkg/ccl/sqlproxyccl/interceptor/interceptor_test.go | |
| index aa1534450d..bb91118dc9 100644 | |
| --- a/pkg/ccl/sqlproxyccl/interceptor/interceptor_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/interceptor/interceptor_test.go | |
| @@ -9,8 +9,7 @@ | |
| package interceptor_test | |
| import ( | |
| - "encoding/json" | |
| - "net" | |
| + "bytes" | |
| "testing" | |
| "github.com/cockroachdb/cockroach/pkg/ccl/sqlproxyccl/interceptor" | |
| @@ -20,100 +19,60 @@ import ( | |
| "github.com/stretchr/testify/require" | |
| ) | |
| -// TestSimpleProxy illustrates how the frontend and backend connections can be | |
| +// TestSimpleProxy illustrates how the frontend and backend interceptors can be | |
| // used as a proxy. | |
| func TestSimpleProxy(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| - t.Run("client to server", func(t *testing.T) { | |
| - // These represents connections for client<->proxy and proxy<->server. | |
| - clientProxy, client := net.Pipe() | |
| - serverProxy, server := net.Pipe() | |
| - defer clientProxy.Close() | |
| - defer client.Close() | |
| - defer serverProxy.Close() | |
| - defer server.Close() | |
| - | |
| - // Create client and server interceptors. | |
| - clientConn := interceptor.NewBackendConn(clientProxy) | |
| - serverConn := interceptor.NewFrontendConn(serverProxy) | |
| + // These represents connections for client<->proxy and proxy<->server. | |
| + fromClient := new(bytes.Buffer) | |
| + toClient := new(bytes.Buffer) | |
| + fromServer := new(bytes.Buffer) | |
| + toServer := new(bytes.Buffer) | |
| + | |
| + // Create client and server interceptors. | |
| + clientInt := interceptor.NewBackendInterceptor(fromClient) | |
| + serverInt := interceptor.NewFrontendInterceptor(fromServer) | |
| + t.Run("client to server", func(t *testing.T) { | |
| // Client sends a list of SQL queries. | |
| queries := []pgproto3.FrontendMessage{ | |
| &pgproto3.Query{String: "SELECT 1"}, | |
| - &pgproto3.Query{String: "SELECT 2 FROM foobar"}, | |
| + &pgproto3.Query{String: "SELECT * FROM foo.bar"}, | |
| &pgproto3.Query{String: "UPDATE foo SET x = 42"}, | |
| &pgproto3.Sync{}, | |
| &pgproto3.Terminate{}, | |
| } | |
| - errCh := make(chan error, len(queries)) | |
| - go func() { | |
| - for _, msg := range queries { | |
| - _, err := client.Write(msg.Encode(nil)) | |
| - errCh <- err | |
| - } | |
| - }() | |
| - msgCh := make(chan pgproto3.FrontendMessage, 10) | |
| - go func() { | |
| - backend := interceptor.NewBackendConn(server) | |
| - for { | |
| - msg, err := backend.ReadMsg() | |
| - if err != nil { | |
| - return | |
| - } | |
| - msgCh <- msg | |
| - } | |
| - }() | |
| + for _, msg := range queries { | |
| + _, err := fromClient.Write(msg.Encode(nil)) | |
| + require.NoError(t, err) | |
| + } | |
| + totalBytes := fromClient.Len() | |
| customQuery := &pgproto3.Query{ | |
| String: "SELECT * FROM crdb_internal.serialize_session()"} | |
| for { | |
| - typ, _, err := clientConn.PeekMsg() | |
| + typ, _, err := clientInt.PeekMsg() | |
| require.NoError(t, err) | |
| // Forward message to server. | |
| - _, err = clientConn.ForwardMsg(serverConn) | |
| + _, err = clientInt.ForwardMsg(toServer) | |
| require.NoError(t, err) | |
| if typ == pgwirebase.ClientMsgTerminate { | |
| // Right before we terminate, we could also craft a custom | |
| // message, and send it to the server. | |
| - _, err := serverConn.Write(customQuery.Encode(nil)) | |
| + _, err := toServer.Write(customQuery.Encode(nil)) | |
| require.NoError(t, err) | |
| break | |
| } | |
| } | |
| - | |
| - expectedMsg := []string{ | |
| - `"Type":"Query","String":"SELECT 1"`, | |
| - `"Type":"Query","String":"SELECT 2 FROM foobar"`, | |
| - `"Type":"Query","String":"UPDATE foo SET x = 42"`, | |
| - `"Type":"Sync"`, | |
| - `"Type":"Terminate"`, | |
| - `"Type":"Query","String":"SELECT \* FROM crdb_internal.serialize_session\(\)"`, | |
| - } | |
| - var m1 []byte | |
| - for _, m2 := range expectedMsg { | |
| - msg := <-msgCh | |
| - m1, _ = json.Marshal(msg) | |
| - require.Regexp(t, m2, string(m1)) | |
| - } | |
| + require.Equal(t, 0, fromClient.Len()) | |
| + require.Equal(t, totalBytes+len(customQuery.Encode(nil)), toServer.Len()) | |
| }) | |
| t.Run("server to client", func(t *testing.T) { | |
| - // These represents connections for client<->proxy and proxy<->server. | |
| - clientProxy, client := net.Pipe() | |
| - serverProxy, server := net.Pipe() | |
| - defer clientProxy.Close() | |
| - defer client.Close() | |
| - defer serverProxy.Close() | |
| - defer server.Close() | |
| - | |
| - // Create client and server interceptors. | |
| - clientConn := interceptor.NewBackendConn(clientProxy) | |
| - serverConn := interceptor.NewFrontendConn(serverProxy) | |
| - | |
| // Server sends back responses. | |
| queries := []pgproto3.BackendMessage{ | |
| // Forward these back to the client. | |
| @@ -124,27 +83,15 @@ func TestSimpleProxy(t *testing.T) { | |
| // Terminator. | |
| &pgproto3.ReadyForQuery{}, | |
| } | |
| - errCh := make(chan error, len(queries)) | |
| - go func() { | |
| - for _, msg := range queries { | |
| - _, err := server.Write(msg.Encode(nil)) | |
| - errCh <- err | |
| - } | |
| - }() | |
| - msgCh := make(chan pgproto3.BackendMessage, 10) | |
| - go func() { | |
| - frontend := interceptor.NewFrontendConn(client) | |
| - for { | |
| - msg, err := frontend.ReadMsg() | |
| - if err != nil { | |
| - return | |
| - } | |
| - msgCh <- msg | |
| - } | |
| - }() | |
| + for _, msg := range queries { | |
| + _, err := fromServer.Write(msg.Encode(nil)) | |
| + require.NoError(t, err) | |
| + } | |
| + // Exclude bytes from second message. | |
| + totalBytes := fromServer.Len() - len(queries[2].Encode(nil)) | |
| for { | |
| - typ, size, err := serverConn.PeekMsg() | |
| + typ, size, err := serverInt.PeekMsg() | |
| require.NoError(t, err) | |
| switch typ { | |
| @@ -152,13 +99,13 @@ func TestSimpleProxy(t *testing.T) { | |
| // Assuming that we're only interested in small messages, then | |
| // we could skip all the large ones. | |
| if size > 12 { | |
| - _, err := serverConn.ForwardMsg(clientConn) | |
| + _, err := serverInt.ForwardMsg(toClient) | |
| require.NoError(t, err) | |
| continue | |
| } | |
| // Decode message. | |
| - msg, err := serverConn.ReadMsg() | |
| + msg, err := serverInt.ReadMsg() | |
| require.NoError(t, err) | |
| // Once we've decoded the message, we could store the message | |
| @@ -167,7 +114,7 @@ func TestSimpleProxy(t *testing.T) { | |
| require.True(t, ok) | |
| require.Equal(t, "short", string(dmsg.CommandTag)) | |
| case pgwirebase.ServerMsgBackendKeyData: | |
| - msg, err := serverConn.ReadMsg() | |
| + msg, err := serverInt.ReadMsg() | |
| require.NoError(t, err) | |
| dmsg, ok := msg.(*pgproto3.BackendKeyData) | |
| @@ -177,11 +124,11 @@ func TestSimpleProxy(t *testing.T) { | |
| // the client. | |
| dmsg.SecretKey = 100 | |
| - _, err = clientConn.Write(dmsg.Encode(nil)) | |
| + _, err = toClient.Write(dmsg.Encode(nil)) | |
| require.NoError(t, err) | |
| default: | |
| // Forward message that we're not interested to the client. | |
| - _, err := serverConn.ForwardMsg(clientConn) | |
| + _, err := serverInt.ForwardMsg(toClient) | |
| require.NoError(t, err) | |
| } | |
| @@ -189,17 +136,7 @@ func TestSimpleProxy(t *testing.T) { | |
| break | |
| } | |
| } | |
| - | |
| - expectedMsg := []string{ | |
| - `"Type":"CommandComplete","CommandTag":"averylongstring"`, | |
| - `"Type":"BackendKeyData","ProcessID":100,"SecretKey":100`, | |
| - `"Type":"ReadyForQuery"`, | |
| - } | |
| - var m1 []byte | |
| - for _, m2 := range expectedMsg { | |
| - msg := <-msgCh | |
| - m1, _ = json.Marshal(msg) | |
| - require.Regexp(t, m2, string(m1)) | |
| - } | |
| + require.Equal(t, 0, fromServer.Len()) | |
| + require.Equal(t, totalBytes, toClient.Len()) | |
| }) | |
| } | |
| diff --git a/pkg/ccl/sqlproxyccl/proxy_handler_test.go b/pkg/ccl/sqlproxyccl/proxy_handler_test.go | |
| index a439d53211..3de0c858c8 100644 | |
| --- a/pkg/ccl/sqlproxyccl/proxy_handler_test.go | |
| +++ b/pkg/ccl/sqlproxyccl/proxy_handler_test.go | |
| @@ -609,9 +609,6 @@ func TestDenylistUpdate(t *testing.T) { | |
| func TestDirectoryConnect(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| - // TODO(jaylim-crl): This is a potential port reuse issue, so skip this | |
| - // under stress. See linked GitHub issue. | |
| - skip.UnderStress(t, "https://github.com/cockroachdb/cockroach/issues/76839") | |
| skip.UnderDeadlockWithIssue(t, 71365) | |
| defer log.Scope(t).Close(t) | |
| diff --git a/pkg/ccl/streamingccl/streamingest/stream_ingestion_planning.go b/pkg/ccl/streamingccl/streamingest/stream_ingestion_planning.go | |
| index 1c738b5ff3..b84076bdfd 100644 | |
| --- a/pkg/ccl/streamingccl/streamingest/stream_ingestion_planning.go | |
| +++ b/pkg/ccl/streamingccl/streamingest/stream_ingestion_planning.go | |
| @@ -121,7 +121,7 @@ func ingestionPlanHook( | |
| } | |
| // We only support a TENANT target, so error out if that is nil. | |
| - if !ingestionStmt.Targets.TenantID.IsSet() { | |
| + if ingestionStmt.Targets.Tenant == (roachpb.TenantID{}) { | |
| return errors.Newf("no tenant specified in ingestion query: %s", ingestionStmt.String()) | |
| } | |
| @@ -154,7 +154,7 @@ func ingestionPlanHook( | |
| // TODO(adityamaru): Add privileges checks. Probably the same as RESTORE. | |
| - prefix := keys.MakeTenantPrefix(ingestionStmt.Targets.TenantID.TenantID) | |
| + prefix := keys.MakeTenantPrefix(ingestionStmt.Targets.Tenant) | |
| startTime := hlc.Timestamp{WallTime: timeutil.Now().UnixNano()} | |
| if ingestionStmt.AsOf.Expr != nil { | |
| asOf, err := p.EvalAsOfTimestamp(ctx, ingestionStmt.AsOf) | |
| @@ -166,7 +166,7 @@ func ingestionPlanHook( | |
| streamIngestionDetails := jobspb.StreamIngestionDetails{ | |
| StreamAddress: string(streamAddress), | |
| - TenantID: ingestionStmt.Targets.TenantID.TenantID, | |
| + TenantID: ingestionStmt.Targets.Tenant, | |
| Span: roachpb.Span{Key: prefix, EndKey: prefix.PrefixEnd()}, | |
| StartTime: startTime, | |
| } | |
| diff --git a/pkg/ccl/streamingccl/streamingest/stream_ingestion_processor.go b/pkg/ccl/streamingccl/streamingest/stream_ingestion_processor.go | |
| index 2d75a64324..94839f51a6 100644 | |
| --- a/pkg/ccl/streamingccl/streamingest/stream_ingestion_processor.go | |
| +++ b/pkg/ccl/streamingccl/streamingest/stream_ingestion_processor.go | |
| @@ -36,7 +36,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/syncutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/logtags" | |
| ) | |
| var minimumFlushInterval = settings.RegisterPublicDurationSettingWithExplicitUnit( | |
| @@ -192,7 +191,6 @@ func newStreamIngestionDataProcessor( | |
| // Start is part of the RowSource interface. | |
| func (sip *streamIngestionProcessor) Start(ctx context.Context) { | |
| - ctx = logtags.AddTag(ctx, "job", sip.spec.JobID) | |
| log.Infof(ctx, "starting ingest proc") | |
| ctx = sip.StartInternal(ctx, streamIngestionProcessorName) | |
| @@ -201,7 +199,8 @@ func (sip *streamIngestionProcessor) Start(ctx context.Context) { | |
| evalCtx := sip.FlowCtx.EvalCtx | |
| db := sip.FlowCtx.Cfg.DB | |
| var err error | |
| - sip.batcher, err = bulk.MakeStreamSSTBatcher(ctx, db, evalCtx.Settings) | |
| + sip.batcher, err = bulk.MakeStreamSSTBatcher(ctx, db, evalCtx.Settings, | |
| + func() int64 { return bulk.IngestFileSize(evalCtx.Settings) }) | |
| if err != nil { | |
| sip.MoveToDraining(errors.Wrap(err, "creating stream sst batcher")) | |
| return | |
| diff --git a/pkg/ccl/streamingccl/streamproducer/replication_stream_planning.go b/pkg/ccl/streamingccl/streamproducer/replication_stream_planning.go | |
| index 82deb3119a..0d44081636 100644 | |
| --- a/pkg/ccl/streamingccl/streamproducer/replication_stream_planning.go | |
| +++ b/pkg/ccl/streamingccl/streamproducer/replication_stream_planning.go | |
| @@ -149,13 +149,13 @@ func doCreateReplicationStream( | |
| } | |
| var spans []roachpb.Span | |
| - if !eval.Targets.TenantID.IsSet() { | |
| + if eval.Targets.Tenant == (roachpb.TenantID{}) { | |
| // TODO(yevgeniy): Only tenant streaming supported now; Support granular streaming. | |
| return pgerror.New(pgcode.FeatureNotSupported, "granular replication streaming not supported") | |
| } | |
| telemetry.Count(`replication.create.tenant`) | |
| - prefix := keys.MakeTenantPrefix(roachpb.MakeTenantID(eval.Targets.TenantID.ToUint64())) | |
| + prefix := keys.MakeTenantPrefix(roachpb.MakeTenantID(eval.Targets.Tenant.ToUint64())) | |
| spans = append(spans, roachpb.Span{ | |
| Key: prefix, | |
| EndKey: prefix.PrefixEnd(), | |
| diff --git a/pkg/ccl/telemetryccl/BUILD.bazel b/pkg/ccl/telemetryccl/BUILD.bazel | |
| index 4ef6c408db..9c1ba639f4 100644 | |
| --- a/pkg/ccl/telemetryccl/BUILD.bazel | |
| +++ b/pkg/ccl/telemetryccl/BUILD.bazel | |
| @@ -7,7 +7,6 @@ go_test( | |
| "telemetry_test.go", | |
| ], | |
| data = glob(["testdata/**"]), | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/ccl", | |
| diff --git a/pkg/ccl/testccl/sqlccl/BUILD.bazel b/pkg/ccl/testccl/sqlccl/BUILD.bazel | |
| index 5f7497c399..a75377363c 100644 | |
| --- a/pkg/ccl/testccl/sqlccl/BUILD.bazel | |
| +++ b/pkg/ccl/testccl/sqlccl/BUILD.bazel | |
| @@ -8,23 +8,15 @@ go_test( | |
| "session_revival_test.go", | |
| "show_transfer_state_test.go", | |
| "temp_table_clean_test.go", | |
| - "tenant_gc_test.go", | |
| ], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/ccl/kvccl/kvtenantccl", | |
| - "//pkg/jobs/jobspb", | |
| - "//pkg/keys", | |
| - "//pkg/roachpb", | |
| "//pkg/security", | |
| "//pkg/security/securitytest", | |
| "//pkg/server", | |
| "//pkg/settings/cluster", | |
| - "//pkg/spanconfig", | |
| "//pkg/sql", | |
| - "//pkg/sql/catalog/descpb", | |
| - "//pkg/sql/gcjob", | |
| "//pkg/sql/sessiondatapb", | |
| "//pkg/sql/sqlliveness/slinstance", | |
| "//pkg/sql/sqltestutils", | |
| diff --git a/pkg/ccl/testccl/sqlccl/session_revival_test.go b/pkg/ccl/testccl/sqlccl/session_revival_test.go | |
| index 97861122ea..83f9a36cab 100644 | |
| --- a/pkg/ccl/testccl/sqlccl/session_revival_test.go | |
| +++ b/pkg/ccl/testccl/sqlccl/session_revival_test.go | |
| @@ -43,9 +43,7 @@ func TestAuthenticateWithSessionRevivalToken(t *testing.T) { | |
| _, err := tenantDB.Exec("CREATE USER testuser WITH PASSWORD 'hunter2'") | |
| require.NoError(t, err) | |
| - // TODO(rafi): use ALTER TENANT ALL when available. | |
| - _, err = mainDB.Exec(`INSERT INTO system.tenant_settings (tenant_id, name, value, value_type) VALUES | |
| - (0, 'server.user_login.session_revival_token.enabled', 'true', 'b')`) | |
| + _, err = tenantDB.Exec("SET CLUSTER SETTING server.user_login.session_revival_token.enabled = true") | |
| require.NoError(t, err) | |
| var token string | |
| diff --git a/pkg/ccl/testccl/sqlccl/show_transfer_state_test.go b/pkg/ccl/testccl/sqlccl/show_transfer_state_test.go | |
| index 7946c7e96d..815951ad99 100644 | |
| --- a/pkg/ccl/testccl/sqlccl/show_transfer_state_test.go | |
| +++ b/pkg/ccl/testccl/sqlccl/show_transfer_state_test.go | |
| @@ -28,17 +28,15 @@ func TestShowTransferState(t *testing.T) { | |
| ctx := context.Background() | |
| params, _ := tests.CreateTestServerParams() | |
| - s, mainDB, _ := serverutils.StartServer(t, params) | |
| + s, _, _ := serverutils.StartServer(t, params) | |
| defer s.Stopper().Stop(ctx) | |
| - tenant, tenantDB := serverutils.StartTenant(t, s, tests.CreateTestTenantParams(serverutils.TestTenantID())) | |
| + tenant, mainDB := serverutils.StartTenant(t, s, tests.CreateTestTenantParams(serverutils.TestTenantID())) | |
| defer tenant.Stopper().Stop(ctx) | |
| - defer tenantDB.Close() | |
| + defer mainDB.Close() | |
| - _, err := tenantDB.Exec("CREATE USER testuser WITH PASSWORD 'hunter2'") | |
| + _, err := mainDB.Exec("CREATE USER testuser WITH PASSWORD 'hunter2'") | |
| require.NoError(t, err) | |
| - // TODO(rafi): use ALTER TENANT ALL when available. | |
| - _, err = mainDB.Exec(`INSERT INTO system.tenant_settings (tenant_id, name, value, value_type) VALUES | |
| - (0, 'server.user_login.session_revival_token.enabled', 'true', 'b')`) | |
| + _, err = mainDB.Exec("SET CLUSTER SETTING server.user_login.session_revival_token.enabled = true") | |
| require.NoError(t, err) | |
| t.Run("without_transfer_key", func(t *testing.T) { | |
| @@ -172,7 +170,7 @@ func TestShowTransferState(t *testing.T) { | |
| t.Run("root_user", func(t *testing.T) { | |
| var key string | |
| var errVal, sessionState, sessionRevivalToken gosql.NullString | |
| - err := tenantDB.QueryRow(`SHOW TRANSFER STATE WITH 'bar'`).Scan(&errVal, &sessionState, &sessionRevivalToken, &key) | |
| + err := mainDB.QueryRow(`SHOW TRANSFER STATE WITH 'bar'`).Scan(&errVal, &sessionState, &sessionRevivalToken, &key) | |
| require.NoError(t, err) | |
| require.True(t, errVal.Valid) | |
| diff --git a/pkg/ccl/testccl/sqlccl/tenant_gc_test.go b/pkg/ccl/testccl/sqlccl/tenant_gc_test.go | |
| deleted file mode 100644 | |
| index 18c474d6e5..0000000000 | |
| --- a/pkg/ccl/testccl/sqlccl/tenant_gc_test.go | |
| +++ /dev/null | |
| @@ -1,121 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Licensed as a CockroachDB Enterprise file under the Cockroach Community | |
| -// License (the "License"); you may not use this file except in compliance with | |
| -// the License. You may obtain a copy of the License at | |
| -// | |
| -// https://github.com/cockroachdb/cockroach/blob/master/licenses/CCL.txt | |
| - | |
| -package sqlccl | |
| - | |
| -import ( | |
| - "context" | |
| - "testing" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/base" | |
| - "github.com/cockroachdb/cockroach/pkg/jobs/jobspb" | |
| - "github.com/cockroachdb/cockroach/pkg/keys" | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig" | |
| - "github.com/cockroachdb/cockroach/pkg/sql" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/catalog/descpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/gcjob" | |
| - "github.com/cockroachdb/cockroach/pkg/testutils/serverutils" | |
| - "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| - "github.com/cockroachdb/cockroach/pkg/util/log" | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -// TestGCTenantRemovesSpanConfigs ensures that GC-ing a tenant removes all | |
| -// span/system span configs installed by it. | |
| -func TestGCTenantRemovesSpanConfigs(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - ctx := context.Background() | |
| - ts, _, _ := serverutils.StartServer(t, base.TestServerArgs{ | |
| - Knobs: base.TestingKnobs{ | |
| - SpanConfig: &spanconfig.TestingKnobs{ | |
| - // Disable the system tenant's reconciliation process so that we can | |
| - // make assertions on the total number of span configurations in the | |
| - // system. | |
| - ManagerDisableJobCreation: true, | |
| - }, | |
| - }, | |
| - }) | |
| - defer ts.Stopper().Stop(ctx) | |
| - execCfg := ts.ExecutorConfig().(sql.ExecutorConfig) | |
| - scKVAccessor := ts.SpanConfigKVAccessor().(spanconfig.KVAccessor) | |
| - | |
| - gcClosure := func(tenID uint64, progress *jobspb.SchemaChangeGCProgress) error { | |
| - return gcjob.TestingGCTenant(ctx, &execCfg, tenID, progress) | |
| - } | |
| - | |
| - tenantID := roachpb.MakeTenantID(10) | |
| - | |
| - tt, err := ts.StartTenant(ctx, base.TestTenantArgs{ | |
| - TenantID: tenantID, | |
| - TestingKnobs: base.TestingKnobs{ | |
| - SpanConfig: &spanconfig.TestingKnobs{ | |
| - // Disable the tenant's span config reconciliation process, we'll | |
| - // instead manually add system span configs via the KVAccessor. | |
| - ManagerDisableJobCreation: true, | |
| - }, | |
| - }, | |
| - }) | |
| - require.NoError(t, err) | |
| - | |
| - tenantKVAccessor := tt.SpanConfigKVAccessor().(spanconfig.KVAccessor) | |
| - // Write a system span config, set by the tenant, targeting its entire | |
| - // keyspace. | |
| - systemTarget, err := spanconfig.MakeTenantKeyspaceTarget(tenantID, tenantID) | |
| - require.NoError(t, err) | |
| - err = tenantKVAccessor.UpdateSpanConfigRecords(ctx, nil /* toDelete */, []spanconfig.Record{ | |
| - { | |
| - Target: spanconfig.MakeTargetFromSystemTarget(systemTarget), | |
| - Config: roachpb.SpanConfig{}, // Doesn't matter | |
| - }, | |
| - }) | |
| - require.NoError(t, err) | |
| - | |
| - // Ensure there are 2 configs for the tenant -- one that spans its entire | |
| - // keyspace, installed on creation, and of course the system span config we | |
| - // inserted above. | |
| - tenPrefix := keys.MakeTenantPrefix(tenantID) | |
| - records, err := tenantKVAccessor.GetSpanConfigRecords(ctx, spanconfig.Targets{ | |
| - spanconfig.MakeTargetFromSpan(roachpb.Span{Key: tenPrefix, EndKey: tenPrefix.PrefixEnd()}), | |
| - spanconfig.MakeTargetFromSystemTarget(systemTarget), | |
| - }) | |
| - require.NoError(t, err) | |
| - require.Equal(t, 2, len(records)) | |
| - | |
| - // Get the entire span config state, from the system tenant's perspective, | |
| - // which we'll use to compare against once the tenant is GC-ed. | |
| - records, err = scKVAccessor.GetSpanConfigRecords( | |
| - ctx, spanconfig.TestingEntireSpanConfigurationStateTargets(), | |
| - ) | |
| - require.NoError(t, err) | |
| - beforeDelete := len(records) | |
| - | |
| - // Mark the tenant as dropped by updating its record. | |
| - require.NoError(t, sql.TestingUpdateTenantRecord( | |
| - ctx, &execCfg, nil, /* txn */ | |
| - &descpb.TenantInfo{ID: tenantID.ToUint64(), State: descpb.TenantInfo_DROP}, | |
| - )) | |
| - | |
| - // Run GC on the tenant. | |
| - progress := &jobspb.SchemaChangeGCProgress{ | |
| - Tenant: &jobspb.SchemaChangeGCProgress_TenantProgress{ | |
| - Status: jobspb.SchemaChangeGCProgress_DELETING, | |
| - }, | |
| - } | |
| - require.NoError(t, gcClosure(tenantID.ToUint64(), progress)) | |
| - require.Equal(t, jobspb.SchemaChangeGCProgress_DELETED, progress.Tenant.Status) | |
| - | |
| - // Ensure the tenant's span configs and system span configs have been deleted. | |
| - records, err = scKVAccessor.GetSpanConfigRecords( | |
| - ctx, spanconfig.TestingEntireSpanConfigurationStateTargets(), | |
| - ) | |
| - require.NoError(t, err) | |
| - require.Equal(t, len(records), beforeDelete-2) | |
| -} | |
| diff --git a/pkg/ccl/workloadccl/allccl/BUILD.bazel b/pkg/ccl/workloadccl/allccl/BUILD.bazel | |
| index 7e9d849184..a2a29ed991 100644 | |
| --- a/pkg/ccl/workloadccl/allccl/BUILD.bazel | |
| +++ b/pkg/ccl/workloadccl/allccl/BUILD.bazel | |
| @@ -40,7 +40,6 @@ go_test( | |
| "main_test.go", | |
| ], | |
| embed = [":allccl"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/ccl", | |
| diff --git a/pkg/cli/debug.go b/pkg/cli/debug.go | |
| index 2797bd981c..9e9c74f52e 100644 | |
| --- a/pkg/cli/debug.go | |
| +++ b/pkg/cli/debug.go | |
| @@ -131,9 +131,8 @@ func parsePositiveDuration(arg string) (time.Duration, error) { | |
| // OpenEngineOptions tunes the behavior of OpenEngine. | |
| type OpenEngineOptions struct { | |
| - ReadOnly bool | |
| - MustExist bool | |
| - DisableAutomaticCompactions bool | |
| + ReadOnly bool | |
| + MustExist bool | |
| } | |
| func (opts OpenEngineOptions) configOptions() []storage.ConfigOption { | |
| @@ -144,22 +143,13 @@ func (opts OpenEngineOptions) configOptions() []storage.ConfigOption { | |
| if opts.MustExist { | |
| cfgOpts = append(cfgOpts, storage.MustExist) | |
| } | |
| - if opts.DisableAutomaticCompactions { | |
| - cfgOpts = append(cfgOpts, storage.DisableAutomaticCompactions) | |
| - } | |
| return cfgOpts | |
| } | |
| -// OpenExistingStore opens the Pebble engine rooted at 'dir'. If 'readOnly' is | |
| -// true, opens the store in read-only mode. If 'disableAutomaticCompactions' is | |
| -// true, disables automatic/background compactions (only used for manual | |
| -// compactions). | |
| -func OpenExistingStore( | |
| - dir string, stopper *stop.Stopper, readOnly, disableAutomaticCompactions bool, | |
| -) (storage.Engine, error) { | |
| - return OpenEngine(dir, stopper, OpenEngineOptions{ | |
| - ReadOnly: readOnly, MustExist: true, DisableAutomaticCompactions: disableAutomaticCompactions, | |
| - }) | |
| +// OpenExistingStore opens the Pebble engine rooted at 'dir'. | |
| +// If 'readOnly' is true, opens the store in read-only mode. | |
| +func OpenExistingStore(dir string, stopper *stop.Stopper, readOnly bool) (storage.Engine, error) { | |
| + return OpenEngine(dir, stopper, OpenEngineOptions{ReadOnly: readOnly, MustExist: true}) | |
| } | |
| // OpenEngine opens the engine at 'dir'. Depending on the supplied options, | |
| @@ -247,7 +237,7 @@ func runDebugKeys(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -420,7 +410,7 @@ func runDebugRangeData(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -510,7 +500,7 @@ func runDebugRangeDescriptors(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -638,7 +628,7 @@ func runDebugRaftLog(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -708,7 +698,7 @@ func runDebugGCCmd(cmd *cobra.Command, args []string) error { | |
| } | |
| } | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -806,7 +796,7 @@ func runDebugCompact(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, false /* readOnly */, true /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, false /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -1074,7 +1064,7 @@ func runDebugUnsafeRemoveDeadReplicas(cmd *cobra.Command, args []string) error { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(args[0], stopper, false /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, false /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -1436,7 +1426,7 @@ func runDebugIntentCount(cmd *cobra.Command, args []string) error { | |
| ctx := context.Background() | |
| defer stopper.Stop(ctx) | |
| - db, err := OpenExistingStore(args[0], stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(args[0], stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| diff --git a/pkg/cli/debug_check_store.go b/pkg/cli/debug_check_store.go | |
| index 44136ff31f..240eb61246 100644 | |
| --- a/pkg/cli/debug_check_store.go | |
| +++ b/pkg/cli/debug_check_store.go | |
| @@ -146,7 +146,7 @@ func checkStoreRangeStats( | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(ctx) | |
| - eng, err := OpenExistingStore(dir, stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + eng, err := OpenExistingStore(dir, stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -220,7 +220,7 @@ func checkStoreRaftState( | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(context.Background()) | |
| - db, err := OpenExistingStore(dir, stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(dir, stopper, true /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| diff --git a/pkg/cli/debug_recover_loss_of_quorum.go b/pkg/cli/debug_recover_loss_of_quorum.go | |
| index ca45a07795..7080edb996 100644 | |
| --- a/pkg/cli/debug_recover_loss_of_quorum.go | |
| +++ b/pkg/cli/debug_recover_loss_of_quorum.go | |
| @@ -203,7 +203,7 @@ func runDebugDeadReplicaCollect(cmd *cobra.Command, args []string) error { | |
| var stores []storage.Engine | |
| for _, storeSpec := range debugRecoverCollectInfoOpts.Stores.Specs { | |
| - db, err := OpenExistingStore(storeSpec.Path, stopper, true /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(storeSpec.Path, stopper, true /* readOnly */) | |
| if err != nil { | |
| return errors.Wrapf(err, "failed to open store at path %q, ensure that store path is "+ | |
| "correct and that it is not used by another process", storeSpec.Path) | |
| @@ -457,7 +457,7 @@ func runDebugExecuteRecoverPlan(cmd *cobra.Command, args []string) error { | |
| var localNodeID roachpb.NodeID | |
| batches := make(map[roachpb.StoreID]storage.Batch) | |
| for _, storeSpec := range debugRecoverExecuteOpts.Stores.Specs { | |
| - store, err := OpenExistingStore(storeSpec.Path, stopper, false /* readOnly */, false /* disableAutomaticCompactions */) | |
| + store, err := OpenExistingStore(storeSpec.Path, stopper, false /* readOnly */) | |
| if err != nil { | |
| return errors.Wrapf(err, "failed to open store at path %q. ensure that store path is "+ | |
| "correct and that it is not used by another process", storeSpec.Path) | |
| diff --git a/pkg/cli/debug_recover_loss_of_quorum_test.go b/pkg/cli/debug_recover_loss_of_quorum_test.go | |
| index d59314b6b6..863738a361 100644 | |
| --- a/pkg/cli/debug_recover_loss_of_quorum_test.go | |
| +++ b/pkg/cli/debug_recover_loss_of_quorum_test.go | |
| @@ -265,52 +265,3 @@ func createIntentOnRangeDescriptor( | |
| t.Fatal(err) | |
| } | |
| } | |
| - | |
| -// TestJsonSerialization verifies that all fields serialized in JSON could be | |
| -// read back. This specific test addresses issues where default naming scheme | |
| -// may not work in combination with other tags correctly. e.g. repeated used | |
| -// with omitempty seem to use camelcase unless explicitly specified. | |
| -func TestJsonSerialization(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - | |
| - rt := roachpb.VOTER_INCOMING | |
| - nr := loqrecoverypb.NodeReplicaInfo{ | |
| - Replicas: []loqrecoverypb.ReplicaInfo{ | |
| - { | |
| - NodeID: 1, | |
| - StoreID: 2, | |
| - Desc: roachpb.RangeDescriptor{ | |
| - RangeID: 3, | |
| - StartKey: roachpb.RKey(keys.MetaMin), | |
| - EndKey: roachpb.RKey(keys.MetaMax), | |
| - InternalReplicas: []roachpb.ReplicaDescriptor{ | |
| - { | |
| - NodeID: 1, | |
| - StoreID: 2, | |
| - ReplicaID: 3, | |
| - Type: &rt, | |
| - }, | |
| - }, | |
| - NextReplicaID: 4, | |
| - Generation: 7, | |
| - }, | |
| - RaftAppliedIndex: 13, | |
| - RaftCommittedIndex: 19, | |
| - RaftLogDescriptorChanges: []loqrecoverypb.DescriptorChangeInfo{ | |
| - { | |
| - ChangeType: 1, | |
| - Desc: &roachpb.RangeDescriptor{}, | |
| - OtherDesc: &roachpb.RangeDescriptor{}, | |
| - }, | |
| - }, | |
| - }, | |
| - }, | |
| - } | |
| - jsonpb := protoutil.JSONPb{Indent: " "} | |
| - data, err := jsonpb.Marshal(nr) | |
| - require.NoError(t, err) | |
| - | |
| - var nrFromJSON loqrecoverypb.NodeReplicaInfo | |
| - require.NoError(t, jsonpb.Unmarshal(data, &nrFromJSON)) | |
| - require.Equal(t, nr, nrFromJSON, "objects before and after serialization") | |
| -} | |
| diff --git a/pkg/cli/debug_test.go b/pkg/cli/debug_test.go | |
| index 56dc90fb8c..b4e6473ae6 100644 | |
| --- a/pkg/cli/debug_test.go | |
| +++ b/pkg/cli/debug_test.go | |
| @@ -80,7 +80,7 @@ func TestOpenExistingStore(t *testing.T) { | |
| }, | |
| } { | |
| t.Run(fmt.Sprintf("dir=%s", test.dir), func(t *testing.T) { | |
| - _, err := OpenExistingStore(test.dir, stopper, false /* readOnly */, false /* disableAutomaticCompactions */) | |
| + _, err := OpenExistingStore(test.dir, stopper, false /* readOnly */) | |
| if !testutils.IsError(err, test.expErr) { | |
| t.Errorf("wanted %s but got %v", test.expErr, err) | |
| } | |
| @@ -114,7 +114,7 @@ func TestOpenReadOnlyStore(t *testing.T) { | |
| }, | |
| } { | |
| t.Run(fmt.Sprintf("readOnly=%t", test.readOnly), func(t *testing.T) { | |
| - db, err := OpenExistingStore(storePath, stopper, test.readOnly, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(storePath, stopper, test.readOnly) | |
| if err != nil { | |
| t.Fatal(err) | |
| } | |
| @@ -255,7 +255,7 @@ func TestRemoveDeadReplicas(t *testing.T) { | |
| stopper := stop.NewStopper() | |
| defer stopper.Stop(ctx) | |
| - db, err := OpenExistingStore(storePaths[idx], stopper, false /* readOnly */, false /* disableAutomaticCompactions */) | |
| + db, err := OpenExistingStore(storePaths[idx], stopper, false /* readOnly */) | |
| if err != nil { | |
| return err | |
| } | |
| diff --git a/pkg/cli/democluster/BUILD.bazel b/pkg/cli/democluster/BUILD.bazel | |
| index d4fe734e33..6877ca2cac 100644 | |
| --- a/pkg/cli/democluster/BUILD.bazel | |
| +++ b/pkg/cli/democluster/BUILD.bazel | |
| @@ -54,7 +54,6 @@ go_test( | |
| name = "democluster_test", | |
| srcs = ["demo_cluster_test.go"], | |
| embed = [":democluster"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/ccl/kvccl/kvtenantccl", | |
| diff --git a/pkg/cli/mt_proxy.go b/pkg/cli/mt_proxy.go | |
| index dff7bbf49a..7db9adbcb5 100644 | |
| --- a/pkg/cli/mt_proxy.go | |
| +++ b/pkg/cli/mt_proxy.go | |
| @@ -23,7 +23,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log/severity" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| "github.com/spf13/cobra" | |
| ) | |
| @@ -148,7 +147,7 @@ func waitForSignals( | |
| } | |
| log.Ops.Shoutf(ctx, severity.ERROR, | |
| - "received signal '%s' during shutdown, initiating hard shutdown", redact.Safe(sig)) | |
| + "received signal '%s' during shutdown, initiating hard shutdown", log.Safe(sig)) | |
| panic("terminate") | |
| case <-stopper.IsStopped(): | |
| const msgDone = "server shutdown completed" | |
| diff --git a/pkg/cli/node.go b/pkg/cli/node.go | |
| index 421142b22f..0b8f72f62e 100644 | |
| --- a/pkg/cli/node.go | |
| +++ b/pkg/cli/node.go | |
| @@ -29,7 +29,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/retry" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| "github.com/spf13/cobra" | |
| "google.golang.org/grpc/codes" | |
| "google.golang.org/grpc/status" | |
| @@ -399,7 +398,7 @@ func handleNodeDecommissionSelf( | |
| cliflags.NodeDecommissionSelf.Name) | |
| } | |
| - log.Infof(ctx, "%s node %d", redact.Safe(command), localNodeID) | |
| + log.Infof(ctx, "%s node %d", log.Safe(command), localNodeID) | |
| return []roachpb.NodeID{localNodeID}, nil | |
| } | |
| diff --git a/pkg/cli/start.go b/pkg/cli/start.go | |
| index 0769d47507..6b5e6a319c 100644 | |
| --- a/pkg/cli/start.go | |
| +++ b/pkg/cli/start.go | |
| @@ -888,7 +888,7 @@ func waitForShutdown( | |
| // shutdown process. | |
| log.Ops.Shoutf(shutdownCtx, severity.ERROR, | |
| "received signal '%s' during shutdown, initiating hard shutdown%s", | |
| - redact.Safe(sig), redact.Safe(hardShutdownHint)) | |
| + log.Safe(sig), log.Safe(hardShutdownHint)) | |
| handleSignalDuringShutdown(sig) | |
| panic("unreachable") | |
| @@ -1187,7 +1187,7 @@ func setupAndInitializeLoggingAndProfiling( | |
| "- %s\n"+ | |
| "- %s", | |
| build.MakeIssueURL(53404), | |
| - redact.Safe(docs.URL("secure-a-cluster.html")), | |
| + log.Safe(docs.URL("secure-a-cluster.html")), | |
| ) | |
| } | |
| @@ -1201,7 +1201,7 @@ func setupAndInitializeLoggingAndProfiling( | |
| "For more information, see:\n\n" + | |
| "- %s" | |
| log.Shoutf(ctx, severity.WARNING, warningString, | |
| - redact.Safe(docs.URL("cockroach-start.html#locality"))) | |
| + log.Safe(docs.URL("cockroach-start.html#locality"))) | |
| } | |
| } | |
| diff --git a/pkg/cli/testdata/zip/partial1 b/pkg/cli/testdata/zip/partial1 | |
| index c9d9ba8891..0e45edfaea 100644 | |
| --- a/pkg/cli/testdata/zip/partial1 | |
| +++ b/pkg/cli/testdata/zip/partial1 | |
| @@ -39,12 +39,8 @@ debug zip --concurrency=1 --cpu-profile-duration=0s /dev/null | |
| [cluster] retrieving SQL data for crdb_internal.invalid_objects... writing output: debug/crdb_internal.invalid_objects.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.index_usage_statistics... writing output: debug/crdb_internal.index_usage_statistics.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... writing output: debug/crdb_internal.table_indexes.txt... done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... writing output: debug/crdb_internal.transaction_contention_events.txt... done | |
| [cluster] requesting nodes... received response... converting to JSON... writing binary output: debug/nodes.json... done | |
| [cluster] requesting liveness... received response... converting to JSON... writing binary output: debug/liveness.json... done | |
| -[cluster] requesting tenant ranges... received response... | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... done | |
| [node 1] node status... converting to JSON... writing binary output: debug/nodes/1/status.json... done | |
| [node 1] using SQL connection URL: postgresql://... | |
| [node 1] retrieving SQL data for crdb_internal.feature_usage... writing output: debug/nodes/1/crdb_internal.feature_usage.txt... done | |
| @@ -300,4 +296,3 @@ debug zip --concurrency=1 --cpu-profile-duration=0s /dev/null | |
| [node 3] writing range 44... converting to JSON... writing binary output: debug/nodes/3/ranges/44.json... done | |
| [cluster] pprof summary script... writing binary output: debug/pprof-summary.sh... done | |
| [cluster] hot range summary script... writing binary output: debug/hot-ranges.sh... done | |
| -[cluster] tenant hot range summary script... writing binary output: debug/hot-ranges-tenant.sh... done | |
| diff --git a/pkg/cli/testdata/zip/partial1_excluded b/pkg/cli/testdata/zip/partial1_excluded | |
| index 73d6fef32c..eefecf24a6 100644 | |
| --- a/pkg/cli/testdata/zip/partial1_excluded | |
| +++ b/pkg/cli/testdata/zip/partial1_excluded | |
| @@ -39,12 +39,8 @@ debug zip /dev/null --concurrency=1 --exclude-nodes=2 --cpu-profile-duration=0 | |
| [cluster] retrieving SQL data for crdb_internal.invalid_objects... writing output: debug/crdb_internal.invalid_objects.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.index_usage_statistics... writing output: debug/crdb_internal.index_usage_statistics.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... writing output: debug/crdb_internal.table_indexes.txt... done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... writing output: debug/crdb_internal.transaction_contention_events.txt... done | |
| [cluster] requesting nodes... received response... converting to JSON... writing binary output: debug/nodes.json... done | |
| [cluster] requesting liveness... received response... converting to JSON... writing binary output: debug/liveness.json... done | |
| -[cluster] requesting tenant ranges... received response... | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... done | |
| [node 1] node status... converting to JSON... writing binary output: debug/nodes/1/status.json... done | |
| [node 1] using SQL connection URL: postgresql://... | |
| [node 1] retrieving SQL data for crdb_internal.feature_usage... writing output: debug/nodes/1/crdb_internal.feature_usage.txt... done | |
| @@ -212,4 +208,3 @@ debug zip /dev/null --concurrency=1 --exclude-nodes=2 --cpu-profile-duration=0 | |
| [node 3] writing range 44... converting to JSON... writing binary output: debug/nodes/3/ranges/44.json... done | |
| [cluster] pprof summary script... writing binary output: debug/pprof-summary.sh... done | |
| [cluster] hot range summary script... writing binary output: debug/hot-ranges.sh... done | |
| -[cluster] tenant hot range summary script... writing binary output: debug/hot-ranges-tenant.sh... done | |
| diff --git a/pkg/cli/testdata/zip/partial2 b/pkg/cli/testdata/zip/partial2 | |
| index 840176f0d3..8ecd45bb24 100644 | |
| --- a/pkg/cli/testdata/zip/partial2 | |
| +++ b/pkg/cli/testdata/zip/partial2 | |
| @@ -39,12 +39,8 @@ debug zip --concurrency=1 --cpu-profile-duration=0 /dev/null | |
| [cluster] retrieving SQL data for crdb_internal.invalid_objects... writing output: debug/crdb_internal.invalid_objects.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.index_usage_statistics... writing output: debug/crdb_internal.index_usage_statistics.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... writing output: debug/crdb_internal.table_indexes.txt... done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... writing output: debug/crdb_internal.transaction_contention_events.txt... done | |
| [cluster] requesting nodes... received response... converting to JSON... writing binary output: debug/nodes.json... done | |
| [cluster] requesting liveness... received response... converting to JSON... writing binary output: debug/liveness.json... done | |
| -[cluster] requesting tenant ranges... received response... | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... done | |
| [node 1] node status... converting to JSON... writing binary output: debug/nodes/1/status.json... done | |
| [node 1] using SQL connection URL: postgresql://... | |
| [node 1] retrieving SQL data for crdb_internal.feature_usage... writing output: debug/nodes/1/crdb_internal.feature_usage.txt... done | |
| @@ -211,4 +207,3 @@ debug zip --concurrency=1 --cpu-profile-duration=0 /dev/null | |
| [node 3] writing range 44... converting to JSON... writing binary output: debug/nodes/3/ranges/44.json... done | |
| [cluster] pprof summary script... writing binary output: debug/pprof-summary.sh... done | |
| [cluster] hot range summary script... writing binary output: debug/hot-ranges.sh... done | |
| -[cluster] tenant hot range summary script... writing binary output: debug/hot-ranges-tenant.sh... done | |
| diff --git a/pkg/cli/testdata/zip/testzip b/pkg/cli/testdata/zip/testzip | |
| index 3001f0fdaa..4692b0a03d 100644 | |
| --- a/pkg/cli/testdata/zip/testzip | |
| +++ b/pkg/cli/testdata/zip/testzip | |
| @@ -39,12 +39,8 @@ debug zip --concurrency=1 --cpu-profile-duration=1s /dev/null | |
| [cluster] retrieving SQL data for crdb_internal.invalid_objects... writing output: debug/crdb_internal.invalid_objects.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.index_usage_statistics... writing output: debug/crdb_internal.index_usage_statistics.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... writing output: debug/crdb_internal.table_indexes.txt... done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... writing output: debug/crdb_internal.transaction_contention_events.txt... done | |
| [cluster] requesting nodes... received response... converting to JSON... writing binary output: debug/nodes.json... done | |
| [cluster] requesting liveness... received response... converting to JSON... writing binary output: debug/liveness.json... done | |
| -[cluster] requesting tenant ranges... received response... | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... done | |
| [cluster] requesting CPU profiles | |
| [cluster] profiles generated | |
| [cluster] profile for node 1... writing binary output: debug/nodes/1/cpu.pprof... done | |
| @@ -129,4 +125,3 @@ debug zip --concurrency=1 --cpu-profile-duration=1s /dev/null | |
| [node 1] writing range 44... converting to JSON... writing binary output: debug/nodes/1/ranges/44.json... done | |
| [cluster] pprof summary script... writing binary output: debug/pprof-summary.sh... done | |
| [cluster] hot range summary script... writing binary output: debug/hot-ranges.sh... done | |
| -[cluster] tenant hot range summary script... writing binary output: debug/hot-ranges-tenant.sh... done | |
| diff --git a/pkg/cli/testdata/zip/testzip_concurrent b/pkg/cli/testdata/zip/testzip_concurrent | |
| index ca38f82986..c3254f4197 100644 | |
| --- a/pkg/cli/testdata/zip/testzip_concurrent | |
| +++ b/pkg/cli/testdata/zip/testzip_concurrent | |
| @@ -40,11 +40,6 @@ zip | |
| [cluster] requesting nodes: done | |
| [cluster] requesting nodes: received response... | |
| [cluster] requesting nodes: writing binary output: debug/nodes.json... | |
| -[cluster] requesting tenant ranges... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... | |
| -[cluster] requesting tenant ranges: done | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: received response... | |
| [cluster] retrieving SQL data for "".crdb_internal.create_schema_statements... | |
| [cluster] retrieving SQL data for "".crdb_internal.create_schema_statements: done | |
| [cluster] retrieving SQL data for "".crdb_internal.create_schema_statements: writing output: debug/crdb_internal.create_schema_statements.txt... | |
| @@ -108,9 +103,6 @@ zip | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes: done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes: writing output: debug/crdb_internal.table_indexes.txt... | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events: done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events: writing output: debug/crdb_internal.transaction_contention_events.txt... | |
| [cluster] retrieving SQL data for crdb_internal.zones... | |
| [cluster] retrieving SQL data for crdb_internal.zones: done | |
| [cluster] retrieving SQL data for crdb_internal.zones: writing output: debug/crdb_internal.zones.txt... | |
| @@ -140,9 +132,6 @@ zip | |
| [cluster] retrieving SQL data for system.settings: writing output: debug/system.settings.txt... | |
| [cluster] retrieving the node status to get the SQL address... | |
| [cluster] retrieving the node status to get the SQL address: ... | |
| -[cluster] tenant hot range summary script... | |
| -[cluster] tenant hot range summary script: done | |
| -[cluster] tenant hot range summary script: writing binary output: debug/hot-ranges-tenant.sh... | |
| [cluster] using SQL address: ... | |
| [cluster] using SQL address: ... | |
| [cluster] using SQL address: ... | |
| diff --git a/pkg/cli/testdata/zip/testzip_tenant b/pkg/cli/testdata/zip/testzip_tenant | |
| index 573876d7b9..88143c02f3 100644 | |
| --- a/pkg/cli/testdata/zip/testzip_tenant | |
| +++ b/pkg/cli/testdata/zip/testzip_tenant | |
| @@ -53,14 +53,10 @@ debug zip --concurrency=1 --cpu-profile-duration=1s /dev/null | |
| [cluster] retrieving SQL data for crdb_internal.invalid_objects... writing output: debug/crdb_internal.invalid_objects.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.index_usage_statistics... writing output: debug/crdb_internal.index_usage_statistics.txt... done | |
| [cluster] retrieving SQL data for crdb_internal.table_indexes... writing output: debug/crdb_internal.table_indexes.txt... done | |
| -[cluster] retrieving SQL data for crdb_internal.transaction_contention_events... writing output: debug/crdb_internal.transaction_contention_events.txt... done | |
| [cluster] requesting nodes... received response... converting to JSON... writing binary output: debug/nodes.json... done | |
| [cluster] requesting liveness... received response... | |
| [cluster] requesting liveness: last request failed: rpc error: ... | |
| [cluster] requesting liveness: creating error output: debug/liveness.json.err.txt... done | |
| -[cluster] requesting tenant ranges... received response... | |
| -[cluster] requesting tenant ranges: last request failed: rpc error: ... | |
| -[cluster] requesting tenant ranges: creating error output: debug/tenant_ranges.err.txt... done | |
| [cluster] requesting CPU profiles | |
| [cluster] profiles generated | |
| [cluster] profile for node 1... writing binary output: debug/nodes/1/cpu.pprof... done | |
| @@ -115,4 +111,3 @@ debug zip --concurrency=1 --cpu-profile-duration=1s /dev/null | |
| [node 1] requesting ranges: creating error output: debug/nodes/1/ranges.err.txt... done | |
| [cluster] pprof summary script... writing binary output: debug/pprof-summary.sh... done | |
| [cluster] hot range summary script... writing binary output: debug/hot-ranges.sh... done | |
| -[cluster] tenant hot range summary script... writing binary output: debug/hot-ranges-tenant.sh... done | |
| diff --git a/pkg/cli/zip.go b/pkg/cli/zip.go | |
| index cbbb30e491..1952c8953b 100644 | |
| --- a/pkg/cli/zip.go | |
| +++ b/pkg/cli/zip.go | |
| @@ -276,7 +276,7 @@ find . -name cpu.pprof -print0 | xargs -0 go tool pprof -tags | |
| } | |
| } | |
| - // A script to summarize the hottest ranges for a storage server's range reports. | |
| + // A script to summarize the hottest ranges. | |
| { | |
| s := zc.clusterPrinter.start("hot range summary script") | |
| if err := z.createRaw(s, debugBase+"/hot-ranges.sh", []byte(`#!/bin/sh | |
| @@ -286,15 +286,6 @@ find . -path './nodes/*/ranges/*.json' -print0 | xargs -0 grep per_second | sort | |
| } | |
| } | |
| - // A script to summarize the hottest ranges for a tenant's range report. | |
| - { | |
| - s := zc.clusterPrinter.start("tenant hot range summary script") | |
| - if err := z.createRaw(s, debugBase+"/hot-ranges-tenant.sh", []byte(`#!/bin/sh | |
| -find . -path './tenant_ranges/*/*.json' -print0 | xargs -0 grep per_second | sort -rhk3 | head -n 20`)); err != nil { | |
| - return err | |
| - } | |
| - } | |
| - | |
| return nil | |
| } | |
| diff --git a/pkg/cli/zip_cluster_wide.go b/pkg/cli/zip_cluster_wide.go | |
| index 70e5eaf3ff..55ac022ebf 100644 | |
| --- a/pkg/cli/zip_cluster_wide.go | |
| +++ b/pkg/cli/zip_cluster_wide.go | |
| @@ -13,7 +13,6 @@ package cli | |
| import ( | |
| "context" | |
| "fmt" | |
| - "sort" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/liveness/livenesspb" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -33,7 +32,6 @@ const ( | |
| schemaPrefix = debugBase + "/schema" | |
| settingsName = debugBase + "/settings" | |
| problemRangesName = reportsPrefix + "/problemranges" | |
| - tenantRangesName = debugBase + "/tenant_ranges" | |
| ) | |
| // makeClusterWideZipRequests defines the zipRequests that are to be | |
| @@ -112,7 +110,6 @@ var debugZipTablesPerCluster = []string{ | |
| "crdb_internal.invalid_objects", | |
| "crdb_internal.index_usage_statistics", | |
| "crdb_internal.table_indexes", | |
| - "crdb_internal.transaction_contention_events", | |
| } | |
| // nodesInfo holds node details pulled from a SQL or storage node. | |
| @@ -192,41 +189,6 @@ func (zc *debugZipContext) collectClusterData( | |
| livenessByNodeID = lresponse.Statuses | |
| } | |
| } | |
| - | |
| - { | |
| - var tenantRanges *serverpb.TenantRangesResponse | |
| - s := zc.clusterPrinter.start("requesting tenant ranges") | |
| - if requestErr := zc.runZipFn(ctx, s, func(ctx context.Context) error { | |
| - var err error | |
| - tenantRanges, err = zc.status.TenantRanges(ctx, &serverpb.TenantRangesRequest{}) | |
| - return err | |
| - }); requestErr != nil { | |
| - if err := zc.z.createError(s, tenantRangesName, requestErr); err != nil { | |
| - return nodesInfo{}, nil, errors.Wrap(err, "fetching tenant ranges") | |
| - } | |
| - } else { | |
| - s.done() | |
| - rangesFound := 0 | |
| - for locality, rangeList := range tenantRanges.RangesByLocality { | |
| - rangesFound += len(rangeList.Ranges) | |
| - sort.Slice(rangeList.Ranges, func(i, j int) bool { | |
| - return rangeList.Ranges[i].RangeID > rangeList.Ranges[j].RangeID | |
| - }) | |
| - sLocality := zc.clusterPrinter.start("writing tenant ranges for locality: %s", locality) | |
| - prefix := fmt.Sprintf("%s/%s", tenantRangesName, locality) | |
| - for _, r := range rangeList.Ranges { | |
| - sRange := zc.clusterPrinter.start("writing tenant range %d", r.RangeID) | |
| - name := fmt.Sprintf("%s/%d", prefix, r.RangeID) | |
| - if err := zc.z.createJSON(sRange, name+".json", r); err != nil { | |
| - return nodesInfo{}, nil, errors.Wrapf(err, "writing tenant range %d for locality %s", r.RangeID, locality) | |
| - } | |
| - } | |
| - sLocality.done() | |
| - } | |
| - zc.clusterPrinter.info("%d tenant ranges found", rangesFound) | |
| - } | |
| - } | |
| - | |
| return ni, livenessByNodeID, nil | |
| } | |
| diff --git a/pkg/cli/zip_tenant_test.go b/pkg/cli/zip_tenant_test.go | |
| index 099fff97a0..7e69de1f1a 100644 | |
| --- a/pkg/cli/zip_tenant_test.go | |
| +++ b/pkg/cli/zip_tenant_test.go | |
| @@ -46,8 +46,6 @@ func TestTenantZip(t *testing.T) { | |
| StoreSpecs: []base.StoreSpec{{ | |
| Path: hostDir, | |
| }}, | |
| - // TODO(abarganier): Switch to secure mode once underlying infra has been | |
| - // updated to support it. See: https://github.com/cockroachdb/cockroach/issues/77173 | |
| Insecure: true, | |
| TenantArgs: &tenantArgs, | |
| }) | |
| diff --git a/pkg/clusterversion/cockroach_versions.go b/pkg/clusterversion/cockroach_versions.go | |
| index 4bedf909ef..ebac3bddc2 100644 | |
| --- a/pkg/clusterversion/cockroach_versions.go | |
| +++ b/pkg/clusterversion/cockroach_versions.go | |
| @@ -305,12 +305,6 @@ const ( | |
| // stop overwriting the LATEST and checkpoint files during backup execution. | |
| // Instead, it writes new files alongside the old in reserved subdirectories. | |
| BackupDoesNotOverwriteLatestAndCheckpoint | |
| - // EnableDeclarativeSchemaChanger is the version where new declarative schema changer | |
| - // can be used to construct schema change plan node. | |
| - EnableDeclarativeSchemaChanger | |
| - | |
| - // RowLevelTTL is the version where we allow row level TTL tables. | |
| - RowLevelTTL | |
| // ************************************************* | |
| // Step (1): Add new versions here. | |
| @@ -506,14 +500,6 @@ var versionsSingleton = keyedVersions{ | |
| Key: BackupDoesNotOverwriteLatestAndCheckpoint, | |
| Version: roachpb.Version{Major: 21, Minor: 2, Internal: 84}, | |
| }, | |
| - { | |
| - Key: EnableDeclarativeSchemaChanger, | |
| - Version: roachpb.Version{Major: 21, Minor: 2, Internal: 86}, | |
| - }, | |
| - { | |
| - Key: RowLevelTTL, | |
| - Version: roachpb.Version{Major: 21, Minor: 2, Internal: 88}, | |
| - }, | |
| // ************************************************* | |
| // Step (2): Add new versions here. | |
| diff --git a/pkg/clusterversion/key_string.go b/pkg/clusterversion/key_string.go | |
| index f7de307fb5..9bee7041fd 100644 | |
| Binary files a/pkg/clusterversion/key_string.go and b/pkg/clusterversion/key_string.go differ | |
| diff --git a/pkg/cmd/bazci/watch.go b/pkg/cmd/bazci/watch.go | |
| index d860be5874..ad6faceda6 100644 | |
| --- a/pkg/cmd/bazci/watch.go | |
| +++ b/pkg/cmd/bazci/watch.go | |
| @@ -198,12 +198,8 @@ func (w watcher) stageBinaryArtifacts() error { | |
| // These targets don't have stable, predictable locations, so | |
| // they have to be hardcoded. | |
| var ext string | |
| - libDir := "lib" | |
| if usingCrossWindowsConfig() { | |
| ext = "dll" | |
| - // NB: the libs end up in the "bin" subdir of libgeos | |
| - // on Windows. | |
| - libDir = "bin" | |
| } else if usingCrossDarwinConfig() { | |
| ext = "dylib" | |
| } else { | |
| @@ -212,8 +208,8 @@ func (w watcher) stageBinaryArtifacts() error { | |
| switch bin { | |
| case "//c-deps:libgeos": | |
| for _, relBinPath := range []string{ | |
| - fmt.Sprintf("c-deps/libgeos/%s/libgeos_c.%s", libDir, ext), | |
| - fmt.Sprintf("c-deps/libgeos/%s/libgeos.%s", libDir, ext), | |
| + fmt.Sprintf("c-deps/libgeos/lib/libgeos_c.%s", ext), | |
| + fmt.Sprintf("c-deps/libgeos/lib/libgeos.%s", ext), | |
| } { | |
| err := w.maybeStageArtifact(w.info.binDir, relBinPath, 0644, finalizePhase, copyContentTo) | |
| if err != nil { | |
| diff --git a/pkg/cmd/compile-build/BUILD.bazel b/pkg/cmd/compile-build/BUILD.bazel | |
| new file mode 100644 | |
| index 0000000000..e4c2887470 | |
| --- /dev/null | |
| +++ b/pkg/cmd/compile-build/BUILD.bazel | |
| @@ -0,0 +1,15 @@ | |
| +load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library") | |
| + | |
| +go_library( | |
| + name = "compile-build_lib", | |
| + srcs = ["main.go"], | |
| + importpath = "github.com/cockroachdb/cockroach/pkg/cmd/compile-build", | |
| + visibility = ["//visibility:private"], | |
| + deps = ["//pkg/release"], | |
| +) | |
| + | |
| +go_binary( | |
| + name = "compile-build", | |
| + embed = [":compile-build_lib"], | |
| + visibility = ["//visibility:public"], | |
| +) | |
| diff --git a/pkg/cmd/compile-build/main.go b/pkg/cmd/compile-build/main.go | |
| new file mode 100644 | |
| index 0000000000..26e6e5bd82 | |
| --- /dev/null | |
| +++ b/pkg/cmd/compile-build/main.go | |
| @@ -0,0 +1,54 @@ | |
| +// Copyright 2020 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package main | |
| + | |
| +import ( | |
| + "flag" | |
| + "go/build" | |
| + "log" | |
| + | |
| + "github.com/cockroachdb/cockroach/pkg/release" | |
| +) | |
| + | |
| +func main() { | |
| + pkg, err := build.Import("github.com/cockroachdb/cockroach", "", build.FindOnly) | |
| + if err != nil { | |
| + log.Fatalf("unable to locate CRDB directory: %s", err) | |
| + } | |
| + | |
| + var compileAll = flag.Bool("all", false, "compile all supported builds (darwin, linux, windows)") | |
| + var buildType = flag.String("buildtype", "release", "compile with a different build type. Default: 'release'. Options: 'development', 'release'") | |
| + flag.Parse() | |
| + | |
| + // We compile just the first supported target unless we explicitly told to | |
| + // cross compile. | |
| + targets := release.SupportedTargets[:1] | |
| + if *compileAll { | |
| + targets = release.SupportedTargets | |
| + } | |
| + opts := []release.MakeReleaseOption{ | |
| + release.WithMakeReleaseOptionEnv("MKRELEASE_BUILDTYPE=" + *buildType), | |
| + } | |
| + | |
| + for _, target := range targets { | |
| + if err := release.MakeRelease( | |
| + target, | |
| + pkg.Dir, | |
| + opts..., | |
| + ); err != nil { | |
| + log.Fatal(err) | |
| + } | |
| + } | |
| + | |
| + if err := release.MakeWorkload(pkg.Dir); err != nil { | |
| + log.Fatal(err) | |
| + } | |
| +} | |
| diff --git a/pkg/cmd/dev/cache.go b/pkg/cmd/dev/cache.go | |
| index ad81cddd23..1186538196 100644 | |
| --- a/pkg/cmd/dev/cache.go | |
| +++ b/pkg/cmd/dev/cache.go | |
| @@ -61,28 +61,26 @@ func (d *dev) cache(cmd *cobra.Command, _ []string) error { | |
| if clean { | |
| return d.cleanCache(ctx) | |
| } | |
| - if down { | |
| - return d.tearDownCache(ctx) | |
| - } | |
| if reset { | |
| // Errors here don't really mean much, we can just ignore them. | |
| err := d.tearDownCache(ctx) | |
| if err != nil { | |
| log.Printf("%v\n", err) | |
| } | |
| - } | |
| - bazelRcLine, err := d.setUpCache(ctx) | |
| - if err != nil { | |
| + bazelRcLine, err := d.setUpCache(ctx) | |
| + if bazelRcLine != "" { | |
| + fmt.Printf("Please add `%s` to your ~/.bazelrc\n", bazelRcLine) | |
| + } | |
| return err | |
| } | |
| - errStr, err := d.checkPresenceInBazelRc(bazelRcLine) | |
| - if err != nil { | |
| - return err | |
| + if down { | |
| + return d.tearDownCache(ctx) | |
| } | |
| - if errStr != "" { | |
| - return fmt.Errorf("%s", errStr) | |
| + bazelRcLine, err := d.setUpCache(ctx) | |
| + if bazelRcLine != "" { | |
| + fmt.Printf("Please add `%s` to your ~/.bazelrc\n", bazelRcLine) | |
| } | |
| - return nil | |
| + return err | |
| } | |
| func bazelRemoteCacheDir() (string, error) { | |
| @@ -136,10 +134,6 @@ func (d *dev) setUpCache(ctx context.Context) (string, error) { | |
| log.Printf("Configuring cache...\n") | |
| - err := d.exec.CommandContextInheritingStdStreams(ctx, "bazel", "build", bazelRemoteTarget) | |
| - if err != nil { | |
| - return "", err | |
| - } | |
| bazelRemoteLoc, err := d.exec.CommandContextSilent(ctx, "bazel", "run", bazelRemoteTarget, "--run_under=//build/bazelutil/whereis") | |
| if err != nil { | |
| return "", err | |
| diff --git a/pkg/cmd/dev/doctor.go b/pkg/cmd/dev/doctor.go | |
| index 12615ba210..e5c2504768 100644 | |
| --- a/pkg/cmd/dev/doctor.go | |
| +++ b/pkg/cmd/dev/doctor.go | |
| @@ -13,10 +13,10 @@ package main | |
| import ( | |
| "context" | |
| "errors" | |
| - "fmt" | |
| "io/ioutil" | |
| "log" | |
| "os" | |
| + osexec "os/exec" | |
| "path/filepath" | |
| "runtime" | |
| "strconv" | |
| @@ -79,7 +79,7 @@ func printStdoutAndErr(stdoutStr string, err error) { | |
| if len(stdoutStr) > 0 { | |
| log.Printf("stdout: %s", stdoutStr) | |
| } | |
| - var cmderr *exec.ExitError | |
| + var cmderr *osexec.ExitError | |
| if errors.As(err, &cmderr) { | |
| stderrStr := strings.TrimSpace(string(cmderr.Stderr)) | |
| if len(stderrStr) > 0 { | |
| @@ -105,13 +105,9 @@ func makeDoctorCmd(runE func(cmd *cobra.Command, args []string) error) *cobra.Co | |
| func (d *dev) doctor(cmd *cobra.Command, _ []string) error { | |
| ctx := cmd.Context() | |
| - failures := []string{} | |
| + success := true | |
| noCache := mustGetFlagBool(cmd, noCacheFlag) | |
| noCacheEnv := d.os.Getenv("DEV_NO_REMOTE_CACHE") | |
| - workspace, err := d.getWorkspace(ctx) | |
| - if err != nil { | |
| - return err | |
| - } | |
| if noCacheEnv != "" { | |
| noCache = true | |
| } | |
| @@ -121,15 +117,16 @@ func (d *dev) doctor(cmd *cobra.Command, _ []string) error { | |
| if runtime.GOOS == "darwin" { | |
| stdout, err := d.exec.CommandContextSilent(ctx, "/usr/bin/xcodebuild", "-version") | |
| if err != nil { | |
| - log.Println("Failed to run `/usr/bin/xcodebuild -version`.") | |
| + success = false | |
| + log.Printf("Failed to run `/usr/bin/xcodebuild -version`.") | |
| stdoutStr := strings.TrimSpace(string(stdout)) | |
| printStdoutAndErr(stdoutStr, err) | |
| - failures = append(failures, `You must have a full installation of XCode to build with Bazel. | |
| + log.Println(`You must have a full installation of XCode to build with Bazel. | |
| A command-line tools instance does not suffice. | |
| Please perform the following steps: | |
| 1. Install XCode from the App Store. | |
| 2. Launch Xcode.app at least once to perform one-time initialization of developer tools. | |
| - 3. Run `+"`xcode-select -switch /Applications/Xcode.app/`.") | |
| + 3. Run ` + "`xcode-select -switch /Applications/Xcode.app/`.") | |
| } | |
| } | |
| @@ -139,24 +136,25 @@ Please perform the following steps: | |
| stdout, err := d.exec.CommandContextSilent(ctx, "cmake", "--version") | |
| stdoutStr := strings.TrimSpace(string(stdout)) | |
| if err != nil { | |
| - log.Println("Failed to run `cmake --version`.") | |
| printStdoutAndErr(stdoutStr, err) | |
| - failures = append(failures, "Failed to run `cmake --version`; do you have it installed?") | |
| + success = false | |
| } else { | |
| versionFields := strings.Split(strings.TrimPrefix(stdoutStr, "cmake version "), ".") | |
| if len(versionFields) < 3 { | |
| - failures = append(failures, fmt.Sprintf("malformed cmake version: %q", stdoutStr)) | |
| + log.Printf("malformed cmake version: %q\n", stdoutStr) | |
| + success = false | |
| } else { | |
| major, majorErr := strconv.Atoi(versionFields[0]) | |
| minor, minorErr := strconv.Atoi(versionFields[1]) | |
| if majorErr != nil || minorErr != nil { | |
| - failures = append(failures, fmt.Sprintf("malformed cmake version: %q", stdoutStr)) | |
| + log.Printf("malformed cmake version: %q\n", stdoutStr) | |
| + success = false | |
| } else if major < cmakeRequiredMajor || minor < cmakeRequiredMinor { | |
| - msg := "cmake is too old, upgrade to 3.20.x+" | |
| + log.Printf("cmake is too old, upgrade to 3.20.x+\n") | |
| if runtime.GOOS == "linux" { | |
| - msg = msg + "\n\t If this is a gceworker you can use ./build/bootstrap/bootstrap-debian.sh to update all tools" | |
| + log.Printf("\t If this is a gceworker you can use ./build/bootstrap/bootstrap-debian.sh to update all tools\n") | |
| } | |
| - failures = append(failures, msg) | |
| + success = false | |
| } | |
| } | |
| } | |
| @@ -181,43 +179,41 @@ Please perform the following steps: | |
| } | |
| // Check whether the build is properly configured to use stamping. | |
| - d.log.Println("doctor: running stamp test") | |
| - failedStampTestMsg := "" | |
| - stdout, err := d.exec.CommandContextSilent(ctx, "bazel", "build", "//build/bazelutil:test_stamping") | |
| - if err != nil { | |
| - failedStampTestMsg = "Failed to run `bazel build //build/bazelutil:test_stamping`" | |
| - log.Println(failedStampTestMsg) | |
| - printStdoutAndErr(string(stdout), err) | |
| + passedStampTest := true | |
| + if _, err := d.exec.CommandContextSilent(ctx, "bazel", "build", "//build/bazelutil:test_stamping"); err != nil { | |
| + passedStampTest = false | |
| } else { | |
| bazelBin, err := d.getBazelBin(ctx) | |
| if err != nil { | |
| return err | |
| } | |
| - testStampingTxt := filepath.Join(bazelBin, "build", "bazelutil", "test_stamping.txt") | |
| - fileContents, err := d.os.ReadFile(testStampingTxt) | |
| + fileContents, err := d.os.ReadFile( | |
| + filepath.Join(bazelBin, "build", "bazelutil", "test_stamping.txt")) | |
| if err != nil { | |
| return err | |
| } | |
| if !strings.Contains(fileContents, "STABLE_BUILD_GIT_BUILD_TYPE") { | |
| - failedStampTestMsg = fmt.Sprintf("Could not find STABLE_BUILD_GIT_TYPE in %s\n", testStampingTxt) | |
| + passedStampTest = false | |
| } | |
| } | |
| - if failedStampTestMsg != "" { | |
| - failedStampTestMsg = failedStampTestMsg + fmt.Sprintf(` | |
| -This may be because your Bazel is not configured to "stamp" built executables. | |
| -Make sure one of the following lines is in the file %s/.bazelrc.user: | |
| -`, workspace) | |
| + workspace, err := d.getWorkspace(ctx) | |
| + if err != nil { | |
| + return err | |
| + } | |
| + if !passedStampTest { | |
| + success = false | |
| + log.Printf(`Your machine is not configured to "stamp" your built executables. | |
| +Please add one of the following to your %s/.bazelrc.user:`, workspace) | |
| if runtime.GOOS == "darwin" && runtime.GOARCH == "amd64" { | |
| - failedStampTestMsg = failedStampTestMsg + " build --config=devdarwinx86_64" | |
| + log.Printf(" build --config=devdarwinx86_64") | |
| } else if runtime.GOOS == "linux" && runtime.GOARCH == "amd64" { | |
| - failedStampTestMsg = failedStampTestMsg + " build --config=dev\n" | |
| - failedStampTestMsg = failedStampTestMsg + " OR \n" | |
| - failedStampTestMsg = failedStampTestMsg + " build --config=crosslinux\n" | |
| - failedStampTestMsg = failedStampTestMsg + "The former will use your host toolchain, while the latter will use the cross-compiler that we use in CI." | |
| + log.Printf(" build --config=dev") | |
| + log.Printf(" OR ") | |
| + log.Printf(" build --config=crosslinux") | |
| + log.Printf("The former will use your host toolchain, while the latter will use the cross-compiler that we use in CI.") | |
| } else { | |
| - failedStampTestMsg = failedStampTestMsg + " build --config=dev" | |
| + log.Printf(" build --config=dev") | |
| } | |
| - failures = append(failures, failedStampTestMsg) | |
| } | |
| if !noCache { | |
| @@ -226,20 +222,19 @@ Make sure one of the following lines is in the file %s/.bazelrc.user: | |
| if err != nil { | |
| return err | |
| } | |
| - msg, err := d.checkPresenceInBazelRc(bazelRcLine) | |
| + homeDir, err := os.UserHomeDir() | |
| if err != nil { | |
| return err | |
| } | |
| - if msg != "" { | |
| - failures = append(failures, msg) | |
| + bazelRcContents, err := d.os.ReadFile(filepath.Join(homeDir, ".bazelrc")) | |
| + if err != nil || !strings.Contains(bazelRcContents, bazelRcLine) { | |
| + log.Printf("Please add the string `%s` to your ~/.bazelrc:\n", bazelRcLine) | |
| + log.Printf(" echo \"%s\" >> ~/.bazelrc", bazelRcLine) | |
| + success = false | |
| } | |
| } | |
| - if len(failures) > 0 { | |
| - log.Printf("doctor: encountered %d errors", len(failures)) | |
| - for _, failure := range failures { | |
| - log.Println(failure) | |
| - } | |
| + if !success { | |
| return errors.New("please address the errors described above and try again") | |
| } | |
| @@ -249,38 +244,3 @@ Make sure one of the following lines is in the file %s/.bazelrc.user: | |
| log.Println("You are ready to build :)") | |
| return nil | |
| } | |
| - | |
| -// checkPresenceInBazelRc checks whether the given line is in ~/.bazelrc. | |
| -// If it is, this function returns an empty string and a nil error. | |
| -// If it isn't, this function returns a non-empty human-readable string describing | |
| -// what the user should do to solve the issue and a nil error. | |
| -// In other failure cases the function returns an empty string and a non-nil error. | |
| -func (d *dev) checkPresenceInBazelRc(expectedBazelRcLine string) (string, error) { | |
| - homeDir, err := os.UserHomeDir() | |
| - if err != nil { | |
| - return "", err | |
| - } | |
| - errString := fmt.Sprintf("Please add the string `%s` to your ~/.bazelrc:\n", expectedBazelRcLine) | |
| - errString = errString + fmt.Sprintf(" echo \"%s\" >> ~/.bazelrc", expectedBazelRcLine) | |
| - | |
| - bazelRcContents, err := d.os.ReadFile(filepath.Join(homeDir, ".bazelrc")) | |
| - if err != nil { | |
| - // The file may not exist; that's OK, but the line definitely is | |
| - // not in the file. | |
| - return errString, nil //nolint:returnerrcheck | |
| - } | |
| - found := false | |
| - for _, line := range strings.Split(bazelRcContents, "\n") { | |
| - if !strings.Contains(line, expectedBazelRcLine) { | |
| - continue | |
| - } | |
| - if strings.HasPrefix(strings.TrimSpace(line), "#") { | |
| - continue | |
| - } | |
| - found = true | |
| - } | |
| - if found { | |
| - return "", nil | |
| - } | |
| - return errString, nil | |
| -} | |
| diff --git a/pkg/cmd/dev/io/exec/exec.go b/pkg/cmd/dev/io/exec/exec.go | |
| index 4e6a53a9db..ee217f927f 100644 | |
| --- a/pkg/cmd/dev/io/exec/exec.go | |
| +++ b/pkg/cmd/dev/io/exec/exec.go | |
| @@ -25,21 +25,6 @@ import ( | |
| "github.com/irfansharif/recorder" | |
| ) | |
| -// ExitError is an error type similar to os.ExitError. The stderr for | |
| -// failed commands is captured as `Stderr`. | |
| -type ExitError struct { | |
| - Stderr []byte | |
| - Inner error | |
| -} | |
| - | |
| -func (e *ExitError) Unwrap() error { | |
| - return e.Inner | |
| -} | |
| - | |
| -func (e *ExitError) Error() string { | |
| - return e.Inner.Error() | |
| -} | |
| - | |
| // Exec is a convenience wrapper around the stdlib os/exec package. It lets us: | |
| // | |
| // (a) mock all instances where we shell out, for tests, and | |
| @@ -238,13 +223,13 @@ func (e *Exec) commandContextImpl( | |
| output, err := e.Next(command, func(outTrace, errTrace io.Writer) (string, error) { | |
| cmd := exec.CommandContext(ctx, name, args...) | |
| - var stdoutBuffer, stderrBuffer bytes.Buffer | |
| + var buffer bytes.Buffer | |
| if silent { | |
| - cmd.Stdout = io.MultiWriter(&stdoutBuffer, outTrace) | |
| - cmd.Stderr = io.MultiWriter(&stderrBuffer, errTrace) | |
| + cmd.Stdout = io.MultiWriter(&buffer, outTrace) | |
| + cmd.Stderr = errTrace | |
| } else { | |
| - cmd.Stdout = io.MultiWriter(e.stdout, &stdoutBuffer, outTrace) | |
| - cmd.Stderr = io.MultiWriter(e.stderr, &stderrBuffer, errTrace) | |
| + cmd.Stdout = io.MultiWriter(e.stdout, &buffer, outTrace) | |
| + cmd.Stderr = io.MultiWriter(e.stderr, errTrace) | |
| } | |
| if stdin != nil { | |
| cmd.Stdin = stdin | |
| @@ -255,9 +240,9 @@ func (e *Exec) commandContextImpl( | |
| return "", err | |
| } | |
| if err := cmd.Wait(); err != nil { | |
| - return "", &ExitError{Inner: err, Stderr: stderrBuffer.Bytes()} | |
| + return "", err | |
| } | |
| - return stdoutBuffer.String(), nil | |
| + return buffer.String(), nil | |
| }) | |
| if err != nil { | |
| diff --git a/pkg/cmd/publish-artifacts/BUILD.bazel b/pkg/cmd/publish-artifacts/BUILD.bazel | |
| index 9f1a9704e7..f6d4a538cb 100644 | |
| --- a/pkg/cmd/publish-artifacts/BUILD.bazel | |
| +++ b/pkg/cmd/publish-artifacts/BUILD.bazel | |
| @@ -7,6 +7,7 @@ go_library( | |
| visibility = ["//visibility:private"], | |
| deps = [ | |
| "//pkg/release", | |
| + "//pkg/util/version", | |
| "@com_github_aws_aws_sdk_go//aws", | |
| "@com_github_aws_aws_sdk_go//aws/session", | |
| "@com_github_aws_aws_sdk_go//service/s3", | |
| diff --git a/pkg/cmd/publish-artifacts/main.go b/pkg/cmd/publish-artifacts/main.go | |
| index 62a506d98a..91d41656bd 100644 | |
| --- a/pkg/cmd/publish-artifacts/main.go | |
| +++ b/pkg/cmd/publish-artifacts/main.go | |
| @@ -11,15 +11,21 @@ | |
| package main | |
| import ( | |
| + "bytes" | |
| "flag" | |
| + "fmt" | |
| + "go/build" | |
| "log" | |
| "os" | |
| + "os/exec" | |
| "path/filepath" | |
| + "strings" | |
| "github.com/aws/aws-sdk-go/aws" | |
| "github.com/aws/aws-sdk-go/aws/session" | |
| "github.com/aws/aws-sdk-go/service/s3" | |
| "github.com/cockroachdb/cockroach/pkg/release" | |
| + "github.com/cockroachdb/cockroach/pkg/util/version" | |
| "github.com/kr/pretty" | |
| ) | |
| @@ -44,8 +50,16 @@ var testableS3 = func() (s3putter, error) { | |
| return s3.New(sess), nil | |
| } | |
| +var isRelease = flag.Bool("release", false, "build in release mode instead of bleeding-edge mode") | |
| var destBucket = flag.String("bucket", "", "override default bucket") | |
| +var ( | |
| + // TODO(tamird,benesch,bdarnell): make "latest" a website-redirect | |
| + // rather than a full key. This means that the actual artifact will no | |
| + // longer be named "-latest". | |
| + latestStr = "latest" | |
| +) | |
| + | |
| func main() { | |
| flag.Parse() | |
| @@ -60,17 +74,34 @@ func main() { | |
| if !ok { | |
| log.Fatalf("VCS branch environment variable %s is not set", teamcityBuildBranchKey) | |
| } | |
| - pkg, err := os.Getwd() | |
| - if err != nil { | |
| - log.Fatalf("unable to locate CRDB directory: %s", err) | |
| - } | |
| - // Make sure the WORKSPACE file is in the current working directory. | |
| - _, err = os.Stat(filepath.Join(pkg, "WORKSPACE")) | |
| + pkg, err := build.Import("github.com/cockroachdb/cockroach", "", build.FindOnly) | |
| if err != nil { | |
| log.Fatalf("unable to locate CRDB directory: %s", err) | |
| } | |
| var versionStr string | |
| + var isStableRelease bool | |
| + if *isRelease { | |
| + ver, err := version.Parse(branch) | |
| + if err != nil { | |
| + log.Fatalf("refusing to build release with invalid version name '%s' (err: %s)", branch, err) | |
| + } | |
| + | |
| + // Prerelease returns anything after the `-` and before metadata. eg: `beta` for `1.0.1-beta+metadata` | |
| + if ver.PreRelease() == "" { | |
| + isStableRelease = true | |
| + } | |
| + versionStr = branch | |
| + } else { | |
| + cmd := exec.Command("git", "rev-parse", "HEAD") | |
| + cmd.Dir = pkg.Dir | |
| + log.Printf("%s %s", cmd.Env, cmd.Args) | |
| + out, err := cmd.Output() | |
| + if err != nil { | |
| + log.Fatalf("%s: out=%q err=%s", cmd.Args, out, err) | |
| + } | |
| + versionStr = string(bytes.TrimSpace(out)) | |
| + } | |
| svc, err := testableS3() | |
| if err != nil { | |
| @@ -80,26 +111,112 @@ func main() { | |
| var bucketName string | |
| if len(*destBucket) > 0 { | |
| bucketName = *destBucket | |
| + } else if *isRelease { | |
| + bucketName = "binaries.cockroachdb.com" | |
| } else { | |
| bucketName = "cockroach" | |
| } | |
| log.Printf("Using S3 bucket: %s", bucketName) | |
| releaseVersionStrs := []string{versionStr} | |
| + // Only build `latest` tarballs for stable releases. | |
| + if isStableRelease { | |
| + releaseVersionStrs = append(releaseVersionStrs, latestStr) | |
| + } | |
| + | |
| + if *isRelease { | |
| + buildArchive(svc, opts{ | |
| + PkgDir: pkg.Dir, | |
| + BucketName: bucketName, | |
| + ReleaseVersionStrs: releaseVersionStrs, | |
| + }) | |
| + } | |
| - for _, platform := range []release.Platform{release.PlatformLinux, release.PlatformMacOS, release.PlatformWindows} { | |
| - var o opts | |
| - o.Platform = platform | |
| - o.ReleaseVersionStrs = releaseVersionStrs | |
| - o.PkgDir = pkg | |
| - o.Branch = branch | |
| - o.VersionStr = versionStr | |
| - o.BucketName = bucketName | |
| - o.Branch = branch | |
| + for _, target := range release.SupportedTargets { | |
| + for i, extraArgs := range []struct { | |
| + goflags string | |
| + suffix string | |
| + tags string | |
| + }{ | |
| + {}, | |
| + // TODO(tamird): re-enable deadlock builds. This really wants its | |
| + // own install suffix; it currently pollutes the normal release | |
| + // build cache. | |
| + // | |
| + // {suffix: ".deadlock", tags: "deadlock"}, | |
| + {suffix: ".race", goflags: "-race"}, | |
| + } { | |
| + var o opts | |
| + o.ReleaseVersionStrs = releaseVersionStrs | |
| + o.PkgDir = pkg.Dir | |
| + o.Branch = branch | |
| + o.VersionStr = versionStr | |
| + o.BucketName = bucketName | |
| + o.Branch = branch | |
| + o.BuildType = target.BuildType | |
| + o.GoFlags = extraArgs.goflags | |
| + o.Suffix = extraArgs.suffix + target.Suffix | |
| + o.Tags = extraArgs.tags | |
| - log.Printf("building %s", pretty.Sprint(o)) | |
| + log.Printf("building %s", pretty.Sprint(o)) | |
| - buildOneCockroach(svc, o) | |
| + // TODO(tamird): build deadlock,race binaries for all targets? | |
| + if i > 0 && (*isRelease || !strings.HasSuffix(o.BuildType, "linux-gnu")) { | |
| + log.Printf("skipping auxiliary build") | |
| + continue | |
| + } | |
| + | |
| + buildOneCockroach(svc, o) | |
| + } | |
| + } | |
| + | |
| + if !*isRelease { | |
| + buildOneWorkload(svc, opts{ | |
| + PkgDir: pkg.Dir, | |
| + BucketName: bucketName, | |
| + Branch: branch, | |
| + VersionStr: versionStr, | |
| + }) | |
| + } | |
| +} | |
| + | |
| +func buildArchive(svc s3putter, o opts) { | |
| + for _, releaseVersionStr := range o.ReleaseVersionStrs { | |
| + archiveBase := fmt.Sprintf("cockroach-%s", releaseVersionStr) | |
| + srcArchive := fmt.Sprintf("%s.%s", archiveBase, "src.tgz") | |
| + cmd := exec.Command( | |
| + "make", | |
| + "archive", | |
| + fmt.Sprintf("ARCHIVE_BASE=%s", archiveBase), | |
| + fmt.Sprintf("ARCHIVE=%s", srcArchive), | |
| + ) | |
| + cmd.Dir = o.PkgDir | |
| + cmd.Stdout = os.Stdout | |
| + cmd.Stderr = os.Stderr | |
| + log.Printf("%s %s", cmd.Env, cmd.Args) | |
| + if err := cmd.Run(); err != nil { | |
| + log.Fatalf("%s: %s", cmd.Args, err) | |
| + } | |
| + | |
| + absoluteSrcArchivePath := filepath.Join(o.PkgDir, srcArchive) | |
| + f, err := os.Open(absoluteSrcArchivePath) | |
| + if err != nil { | |
| + log.Fatalf("os.Open(%s): %s", absoluteSrcArchivePath, err) | |
| + } | |
| + putObjectInput := s3.PutObjectInput{ | |
| + Bucket: &o.BucketName, | |
| + Key: &srcArchive, | |
| + Body: f, | |
| + } | |
| + if releaseVersionStr == latestStr { | |
| + putObjectInput.CacheControl = &release.NoCache | |
| + } | |
| + if _, err := svc.PutObject(&putObjectInput); err != nil { | |
| + log.Fatalf("s3 upload %s: %s", absoluteSrcArchivePath, err) | |
| + } | |
| + if err := f.Close(); err != nil { | |
| + log.Fatal(err) | |
| + } | |
| } | |
| } | |
| @@ -109,11 +226,52 @@ func buildOneCockroach(svc s3putter, o opts) { | |
| log.Printf("done building cockroach: %s", pretty.Sprint(o)) | |
| }() | |
| - if err := release.MakeRelease(o.Platform, release.BuildOptions{}, o.PkgDir); err != nil { | |
| + opts := []release.MakeReleaseOption{ | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "GOFLAGS", o.GoFlags)), | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "TAGS", o.Tags)), | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "BUILDCHANNEL", "official-binary")), | |
| + } | |
| + if *isRelease { | |
| + opts = append(opts, release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "BUILD_TAGGED_RELEASE", "true"))) | |
| + } | |
| + | |
| + if err := release.MakeRelease( | |
| + release.SupportedTarget{ | |
| + BuildType: o.BuildType, | |
| + Suffix: o.Suffix, | |
| + }, | |
| + o.PkgDir, | |
| + opts..., | |
| + ); err != nil { | |
| + log.Fatal(err) | |
| + } | |
| + | |
| + o.Base = "cockroach" + o.Suffix | |
| + o.AbsolutePath = filepath.Join(o.PkgDir, o.Base) | |
| + | |
| + if !*isRelease { | |
| + putNonRelease(svc, o, release.MakeCRDBLibraryNonReleaseFiles(o.PkgDir, o.BuildType, o.VersionStr, o.Suffix)...) | |
| + } else { | |
| + putRelease(svc, o) | |
| + } | |
| +} | |
| + | |
| +func buildOneWorkload(svc s3putter, o opts) { | |
| + defer func() { | |
| + log.Printf("done building workload: %s", pretty.Sprint(o)) | |
| + }() | |
| + | |
| + if *isRelease { | |
| + log.Fatalf("refusing to build workload in release mode") | |
| + } | |
| + | |
| + if err := release.MakeWorkload(o.PkgDir); err != nil { | |
| log.Fatal(err) | |
| } | |
| - putNonRelease(svc, o, release.MakeCRDBLibraryNonReleaseFiles(o.PkgDir, o.Platform, o.VersionStr)...) | |
| + o.Base = "workload" | |
| + o.AbsolutePath = filepath.Join(o.PkgDir, "bin", o.Base) | |
| + putNonRelease(svc, o) | |
| } | |
| type opts struct { | |
| @@ -121,8 +279,12 @@ type opts struct { | |
| Branch string | |
| ReleaseVersionStrs []string | |
| - Platform release.Platform | |
| + BuildType string | |
| + GoFlags string | |
| + Suffix string | |
| + Tags string | |
| + Base string | |
| BucketName string | |
| AbsolutePath string | |
| PkgDir string | |
| @@ -135,9 +297,25 @@ func putNonRelease(svc s3putter, o opts, additionalNonReleaseFiles ...release.No | |
| Branch: o.Branch, | |
| BucketName: o.BucketName, | |
| Files: append( | |
| - []release.NonReleaseFile{release.MakeCRDBBinaryNonReleaseFile(o.AbsolutePath, o.VersionStr)}, | |
| + []release.NonReleaseFile{release.MakeCRDBBinaryNonReleaseFile(o.Base, o.AbsolutePath, o.VersionStr)}, | |
| additionalNonReleaseFiles..., | |
| ), | |
| }, | |
| ) | |
| } | |
| + | |
| +func putRelease(svc s3putter, o opts) { | |
| + for _, releaseVersionStr := range o.ReleaseVersionStrs { | |
| + release.PutRelease(svc, release.PutReleaseOptions{ | |
| + BucketName: o.BucketName, | |
| + NoCache: releaseVersionStr == latestStr, | |
| + Suffix: o.Suffix, | |
| + BuildType: o.BuildType, | |
| + VersionStr: releaseVersionStr, | |
| + Files: append( | |
| + []release.ArchiveFile{release.MakeCRDBBinaryArchiveFile(o.Base, o.AbsolutePath)}, | |
| + release.MakeCRDBLibraryArchiveFiles(o.PkgDir, o.BuildType)..., | |
| + ), | |
| + }) | |
| + } | |
| +} | |
| diff --git a/pkg/cmd/publish-artifacts/main_test.go b/pkg/cmd/publish-artifacts/main_test.go | |
| index 351361136d..26e51394f3 100644 | |
| --- a/pkg/cmd/publish-artifacts/main_test.go | |
| +++ b/pkg/cmd/publish-artifacts/main_test.go | |
| @@ -219,6 +219,42 @@ func TestMainF(t *testing.T) { | |
| Key: "cockroach/workload.LATEST", | |
| WebsiteRedirectLocation: "/cockroach/workload." + shaStub, | |
| }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + Key: "cockroach-v42.42.42.src.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + CacheControl: "no-cache", | |
| + Key: "cockroach-latest.src.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + Key: "cockroach-v42.42.42.darwin-10.9-amd64.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + CacheControl: "no-cache", | |
| + Key: "cockroach-latest.darwin-10.9-amd64.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + Key: "cockroach-v42.42.42.linux-amd64.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + CacheControl: "no-cache", | |
| + Key: "cockroach-latest.linux-amd64.tgz", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + Key: "cockroach-v42.42.42.windows-6.2-amd64.zip", | |
| + }, | |
| + { | |
| + Bucket: "binaries.cockroachdb.com", | |
| + CacheControl: "no-cache", | |
| + Key: "cockroach-latest.windows-6.2-amd64.zip", | |
| + }, | |
| } | |
| if err := os.Setenv("TC_BUILD_BRANCH", "master"); err != nil { | |
| @@ -226,6 +262,12 @@ func TestMainF(t *testing.T) { | |
| } | |
| main() | |
| + if err := os.Setenv("TC_BUILD_BRANCH", "v42.42.42"); err != nil { | |
| + t.Fatal(err) | |
| + } | |
| + *isRelease = true | |
| + main() | |
| + | |
| var acts []testCase | |
| for _, req := range r.reqs { | |
| var act testCase | |
| diff --git a/pkg/cmd/publish-provisional-artifacts/BUILD.bazel b/pkg/cmd/publish-provisional-artifacts/BUILD.bazel | |
| index a23da639b1..3d2ab9748c 100644 | |
| --- a/pkg/cmd/publish-provisional-artifacts/BUILD.bazel | |
| +++ b/pkg/cmd/publish-provisional-artifacts/BUILD.bazel | |
| @@ -29,7 +29,6 @@ go_test( | |
| deps = [ | |
| "//pkg/release", | |
| "//pkg/testutils", | |
| - "@com_github_alessio_shellescape//:shellescape", | |
| "@com_github_aws_aws_sdk_go//service/s3", | |
| "@com_github_cockroachdb_errors//:errors", | |
| "@com_github_stretchr_testify//require", | |
| diff --git a/pkg/cmd/publish-provisional-artifacts/main.go b/pkg/cmd/publish-provisional-artifacts/main.go | |
| index d5e79cc7af..bf606560bf 100644 | |
| --- a/pkg/cmd/publish-provisional-artifacts/main.go | |
| +++ b/pkg/cmd/publish-provisional-artifacts/main.go | |
| @@ -13,12 +13,15 @@ package main | |
| import ( | |
| "bytes" | |
| "flag" | |
| + "fmt" | |
| + "go/build" | |
| "io" | |
| "log" | |
| "os" | |
| "os/exec" | |
| "path/filepath" | |
| "regexp" | |
| + "strings" | |
| "github.com/aws/aws-sdk-go/aws" | |
| "github.com/aws/aws-sdk-go/aws/session" | |
| @@ -77,36 +80,31 @@ func main() { | |
| if err != nil { | |
| log.Fatalf("Creating AWS S3 session: %s", err) | |
| } | |
| + execFn := release.DefaultExecFn | |
| branch, ok := os.LookupEnv(teamcityBuildBranchKey) | |
| if !ok { | |
| log.Fatalf("VCS branch environment variable %s is not set", teamcityBuildBranchKey) | |
| } | |
| - pkg, err := os.Getwd() | |
| + pkg, err := build.Import("github.com/cockroachdb/cockroach", "", build.FindOnly) | |
| if err != nil { | |
| log.Fatalf("unable to locate CRDB directory: %s", err) | |
| } | |
| - // Make sure the WORKSPACE file is in the current working directory. | |
| - _, err = os.Stat(filepath.Join(pkg, "WORKSPACE")) | |
| - if err != nil { | |
| - log.Fatalf("unable to locate CRDB directory: %s", err) | |
| - } | |
| - | |
| cmd := exec.Command("git", "rev-parse", "HEAD") | |
| - cmd.Dir = pkg | |
| + cmd.Dir = pkg.Dir | |
| log.Printf("%s %s", cmd.Env, cmd.Args) | |
| shaOut, err := cmd.Output() | |
| if err != nil { | |
| log.Fatalf("%s: out=%q err=%s", cmd.Args, shaOut, err) | |
| } | |
| - run(s3, runFlags{ | |
| + run(s3, execFn, runFlags{ | |
| doProvisional: *doProvisionalF, | |
| doBless: *doBlessF, | |
| isRelease: *isReleaseF, | |
| branch: branch, | |
| - pkgDir: pkg, | |
| + pkgDir: pkg.Dir, | |
| sha: string(bytes.TrimSpace(shaOut)), | |
| - }, release.ExecFn{}) | |
| + }) | |
| } | |
| type runFlags struct { | |
| @@ -116,7 +114,7 @@ type runFlags struct { | |
| pkgDir string | |
| } | |
| -func run(svc s3I, flags runFlags, execFn release.ExecFn) { | |
| +func run(svc s3I, execFn release.ExecFn, flags runFlags) { | |
| // TODO(dan): non-release builds currently aren't broken into the two | |
| // phases. Instead, the provisional phase does them both. | |
| if !flags.isRelease { | |
| @@ -165,21 +163,51 @@ func run(svc s3I, flags runFlags, execFn release.ExecFn) { | |
| log.Printf("Using S3 bucket: %s", bucketName) | |
| var cockroachBuildOpts []opts | |
| - for _, platform := range []release.Platform{release.PlatformLinux, release.PlatformMacOS, release.PlatformWindows} { | |
| - var o opts | |
| - o.Platform = platform | |
| - o.PkgDir = flags.pkgDir | |
| - o.Branch = flags.branch | |
| - o.VersionStr = versionStr | |
| - o.BucketName = bucketName | |
| - o.AbsolutePath = filepath.Join(flags.pkgDir, "cockroach"+release.SuffixFromPlatform(platform)) | |
| - | |
| - cockroachBuildOpts = append(cockroachBuildOpts, o) | |
| + for _, target := range release.SupportedTargets { | |
| + for i, extraArgs := range []struct { | |
| + goflags string | |
| + suffix string | |
| + tags string | |
| + }{ | |
| + {}, | |
| + // TODO(tamird): re-enable deadlock builds. This really wants its | |
| + // own install suffix; it currently pollutes the normal release | |
| + // build cache. | |
| + // | |
| + // {suffix: ".deadlock", tags: "deadlock"}, | |
| + // | |
| + // As of 2019-01-08, we don't really use this for anything, so save the build time. | |
| + // {suffix: ".race", goflags: "-race"}, | |
| + } { | |
| + var o opts | |
| + o.PkgDir = flags.pkgDir | |
| + o.Branch = flags.branch | |
| + o.VersionStr = versionStr | |
| + o.BucketName = bucketName | |
| + o.BuildType = target.BuildType | |
| + o.GoFlags = extraArgs.goflags | |
| + o.Suffix = extraArgs.suffix + target.Suffix | |
| + o.Tags = extraArgs.tags | |
| + o.Base = "cockroach" + o.Suffix | |
| + | |
| + // TODO(tamird): build deadlock,race binaries for all targets? | |
| + if i > 0 && (flags.isRelease || !strings.HasSuffix(o.BuildType, "linux-gnu")) { | |
| + log.Printf("skipping auxiliary build: %s", pretty.Sprint(o)) | |
| + continue | |
| + } | |
| + | |
| + cockroachBuildOpts = append(cockroachBuildOpts, o) | |
| + } | |
| + } | |
| + archiveBuildOpts := opts{ | |
| + PkgDir: flags.pkgDir, | |
| + BucketName: bucketName, | |
| + VersionStr: versionStr, | |
| } | |
| if flags.doProvisional { | |
| for _, o := range cockroachBuildOpts { | |
| - buildCockroach(flags, o, execFn) | |
| + buildCockroach(execFn, flags, o) | |
| if !flags.isRelease { | |
| putNonRelease(svc, o) | |
| @@ -187,6 +215,9 @@ func run(svc s3I, flags runFlags, execFn release.ExecFn) { | |
| putRelease(svc, o) | |
| } | |
| } | |
| + if flags.isRelease { | |
| + buildAndPutArchive(svc, execFn, archiveBuildOpts) | |
| + } | |
| } | |
| if flags.doBless { | |
| if !flags.isRelease { | |
| @@ -196,24 +227,76 @@ func run(svc s3I, flags runFlags, execFn release.ExecFn) { | |
| for _, o := range cockroachBuildOpts { | |
| markLatestRelease(svc, o) | |
| } | |
| + markLatestArchive(svc, archiveBuildOpts) | |
| } | |
| } | |
| } | |
| -func buildCockroach(flags runFlags, o opts, execFn release.ExecFn) { | |
| +func buildAndPutArchive(svc s3I, execFn release.ExecFn, o opts) { | |
| + log.Printf("building archive %s", pretty.Sprint(o)) | |
| + defer func() { | |
| + log.Printf("done building archive: %s", pretty.Sprint(o)) | |
| + }() | |
| + | |
| + archiveBase, srcArchive := s3KeyArchive(o) | |
| + cmd := exec.Command( | |
| + "make", | |
| + "archive", | |
| + fmt.Sprintf("ARCHIVE_BASE=%s", archiveBase), | |
| + fmt.Sprintf("ARCHIVE=%s", srcArchive), | |
| + fmt.Sprintf("BUILDINFO_TAG=%s", o.VersionStr), | |
| + ) | |
| + cmd.Dir = o.PkgDir | |
| + cmd.Stderr = os.Stderr | |
| + log.Printf("%s %s", cmd.Env, cmd.Args) | |
| + if out, err := execFn(cmd); err != nil { | |
| + log.Fatalf("%s %s: %s\n\n%s", cmd.Env, cmd.Args, err, out) | |
| + } | |
| + | |
| + absoluteSrcArchivePath := filepath.Join(o.PkgDir, srcArchive) | |
| + f, err := os.Open(absoluteSrcArchivePath) | |
| + if err != nil { | |
| + log.Fatalf("os.Open(%s): %s", absoluteSrcArchivePath, err) | |
| + } | |
| + log.Printf("Uploading to s3://%s/%s", o.BucketName, srcArchive) | |
| + putObjectInput := s3.PutObjectInput{ | |
| + Bucket: &o.BucketName, | |
| + Key: &srcArchive, | |
| + Body: f, | |
| + } | |
| + if _, err := svc.PutObject(&putObjectInput); err != nil { | |
| + log.Fatalf("s3 upload %s: %s", absoluteSrcArchivePath, err) | |
| + } | |
| + if err := f.Close(); err != nil { | |
| + log.Fatal(err) | |
| + } | |
| +} | |
| + | |
| +func buildCockroach(execFn release.ExecFn, flags runFlags, o opts) { | |
| log.Printf("building cockroach %s", pretty.Sprint(o)) | |
| defer func() { | |
| log.Printf("done building cockroach: %s", pretty.Sprint(o)) | |
| }() | |
| - var buildOpts release.BuildOptions | |
| - buildOpts.ExecFn = execFn | |
| + opts := []release.MakeReleaseOption{ | |
| + release.WithMakeReleaseOptionExecFn(execFn), | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "GOFLAGS", o.GoFlags)), | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "TAGS", o.Tags)), | |
| + release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "BUILDCHANNEL", "official-binary")), | |
| + } | |
| if flags.isRelease { | |
| - buildOpts.Release = true | |
| - buildOpts.BuildTag = o.VersionStr | |
| + opts = append(opts, release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "BUILDINFO_TAG", o.VersionStr))) | |
| + opts = append(opts, release.WithMakeReleaseOptionBuildArg(fmt.Sprintf("%s=%s", "BUILD_TAGGED_RELEASE", "true"))) | |
| } | |
| - if err := release.MakeRelease(o.Platform, buildOpts, o.PkgDir); err != nil { | |
| + if err := release.MakeRelease( | |
| + release.SupportedTarget{ | |
| + BuildType: o.BuildType, | |
| + Suffix: o.Suffix, | |
| + }, | |
| + o.PkgDir, | |
| + opts..., | |
| + ); err != nil { | |
| log.Fatal(err) | |
| } | |
| } | |
| @@ -222,11 +305,14 @@ type opts struct { | |
| VersionStr string | |
| Branch string | |
| - Platform release.Platform | |
| + BuildType string | |
| + GoFlags string | |
| + Suffix string | |
| + Tags string | |
| - AbsolutePath string | |
| - BucketName string | |
| - PkgDir string | |
| + Base string | |
| + BucketName string | |
| + PkgDir string | |
| } | |
| func putNonRelease(svc s3I, o opts) { | |
| @@ -236,26 +322,33 @@ func putNonRelease(svc s3I, o opts) { | |
| Branch: o.Branch, | |
| BucketName: o.BucketName, | |
| Files: append( | |
| - []release.NonReleaseFile{release.MakeCRDBBinaryNonReleaseFile(o.AbsolutePath, o.VersionStr)}, | |
| - release.MakeCRDBLibraryNonReleaseFiles(o.PkgDir, o.Platform, o.VersionStr)..., | |
| + []release.NonReleaseFile{release.MakeCRDBBinaryNonReleaseFile(o.Base, filepath.Join(o.PkgDir, o.Base), o.VersionStr)}, | |
| + release.MakeCRDBLibraryNonReleaseFiles(o.PkgDir, o.BuildType, o.VersionStr, o.Suffix)..., | |
| ), | |
| }, | |
| ) | |
| } | |
| func s3KeyRelease(o opts) (string, string) { | |
| - return release.S3KeyRelease(o.Platform, o.VersionStr) | |
| + return release.S3KeyRelease(o.Suffix, o.BuildType, o.VersionStr) | |
| +} | |
| + | |
| +func s3KeyArchive(o opts) (string, string) { | |
| + archiveBase := fmt.Sprintf("cockroach-%s", o.VersionStr) | |
| + srcArchive := fmt.Sprintf("%s.%s", archiveBase, "src.tgz") | |
| + return archiveBase, srcArchive | |
| } | |
| func putRelease(svc s3I, o opts) { | |
| release.PutRelease(svc, release.PutReleaseOptions{ | |
| BucketName: o.BucketName, | |
| NoCache: false, | |
| - Platform: o.Platform, | |
| + Suffix: o.Suffix, | |
| + BuildType: o.BuildType, | |
| VersionStr: o.VersionStr, | |
| Files: append( | |
| - []release.ArchiveFile{release.MakeCRDBBinaryArchiveFile(o.AbsolutePath)}, | |
| - release.MakeCRDBLibraryArchiveFiles(o.PkgDir, o.Platform)..., | |
| + []release.ArchiveFile{release.MakeCRDBBinaryArchiveFile(o.Base, filepath.Join(o.PkgDir, o.Base))}, | |
| + release.MakeCRDBLibraryArchiveFiles(o.PkgDir, o.BuildType)..., | |
| ), | |
| }) | |
| } | |
| @@ -290,3 +383,35 @@ func markLatestRelease(svc s3I, o opts) { | |
| log.Fatalf("s3 upload %s: %s", keyLatest, err) | |
| } | |
| } | |
| + | |
| +func markLatestArchive(svc s3I, o opts) { | |
| + _, keyRelease := s3KeyArchive(o) | |
| + | |
| + log.Printf("Downloading from %s/%s", o.BucketName, keyRelease) | |
| + binary, err := svc.GetObject(&s3.GetObjectInput{ | |
| + Bucket: &o.BucketName, | |
| + Key: &keyRelease, | |
| + }) | |
| + if err != nil { | |
| + log.Fatal(err) | |
| + } | |
| + defer binary.Body.Close() | |
| + var buf bytes.Buffer | |
| + if _, err := io.Copy(&buf, binary.Body); err != nil { | |
| + log.Fatalf("downloading %s/%s: %s", o.BucketName, keyRelease, err) | |
| + } | |
| + | |
| + oLatest := o | |
| + oLatest.VersionStr = latestStr | |
| + _, keyLatest := s3KeyArchive(oLatest) | |
| + log.Printf("Uploading to s3://%s/%s", o.BucketName, keyLatest) | |
| + putObjectInput := s3.PutObjectInput{ | |
| + Bucket: &o.BucketName, | |
| + Key: &keyLatest, | |
| + Body: bytes.NewReader(buf.Bytes()), | |
| + CacheControl: &release.NoCache, | |
| + } | |
| + if _, err := svc.PutObject(&putObjectInput); err != nil { | |
| + log.Fatalf("s3 upload %s: %s", keyLatest, err) | |
| + } | |
| +} | |
| diff --git a/pkg/cmd/publish-provisional-artifacts/main_test.go b/pkg/cmd/publish-provisional-artifacts/main_test.go | |
| index 4393826ada..a7ebb36c8d 100644 | |
| --- a/pkg/cmd/publish-provisional-artifacts/main_test.go | |
| +++ b/pkg/cmd/publish-provisional-artifacts/main_test.go | |
| @@ -21,7 +21,6 @@ import ( | |
| "testing" | |
| "unicode/utf8" | |
| - "github.com/alessio/shellescape" | |
| "github.com/aws/aws-sdk-go/service/s3" | |
| "github.com/cockroachdb/cockroach/pkg/release" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| @@ -67,64 +66,49 @@ func (s *mockS3) PutObject(i *s3.PutObjectInput) (*s3.PutObjectOutput, error) { | |
| } | |
| type mockExecRunner struct { | |
| - fakeBazelBin string | |
| - cmds []string | |
| + cmds []string | |
| } | |
| func (r *mockExecRunner) run(c *exec.Cmd) ([]byte, error) { | |
| - if r.fakeBazelBin == "" { | |
| - panic("r.fakeBazelBin not set") | |
| - } | |
| if c.Dir == `` { | |
| return nil, errors.Errorf(`Dir must be specified`) | |
| } | |
| - cmd := fmt.Sprintf("env=%s args=%s", c.Env, shellescape.QuoteCommand(c.Args)) | |
| - r.cmds = append(r.cmds, cmd) | |
| + cmd := fmt.Sprintf("env=%s args=%s", c.Env, c.Args) | |
| var paths []string | |
| - if c.Args[0] == "bazel" && c.Args[1] == "info" && c.Args[2] == "bazel-bin" { | |
| - return []byte(r.fakeBazelBin), nil | |
| - } | |
| - if c.Args[0] == "bazel" && c.Args[1] == "build" { | |
| - path := filepath.Join(r.fakeBazelBin, "pkg", "cmd", "cockroach", "cockroach_", "cockroach") | |
| - var platform release.Platform | |
| + if c.Args[0] == `mkrelease` { | |
| + path := filepath.Join(c.Dir, `cockroach`) | |
| for _, arg := range c.Args { | |
| - if strings.HasPrefix(arg, `--config=`) { | |
| - switch strings.TrimPrefix(arg, `--config=`) { | |
| - case "crosslinuxbase": | |
| - platform = release.PlatformLinux | |
| - case "crosslinuxarmbase": | |
| - platform = release.PlatformLinuxArm | |
| - case "crossmacosbase": | |
| - platform = release.PlatformMacOS | |
| - case "crosswindowsbase": | |
| - platform = release.PlatformWindows | |
| - path += ".exe" | |
| - case "ci", "with_ui": | |
| - default: | |
| - panic(fmt.Sprintf("Unexpected configuration %s", arg)) | |
| - } | |
| + if strings.HasPrefix(arg, `SUFFIX=`) { | |
| + path += strings.TrimPrefix(arg, `SUFFIX=`) | |
| } | |
| } | |
| paths = append(paths, path) | |
| - ext := release.SharedLibraryExtensionFromPlatform(platform) | |
| + ext := release.SharedLibraryExtensionFromBuildType(c.Args[1]) | |
| for _, lib := range release.CRDBSharedLibraries { | |
| - libDir := "lib" | |
| - if platform == release.PlatformWindows { | |
| - libDir = "bin" | |
| + paths = append(paths, filepath.Join(c.Dir, "lib", lib+ext)) | |
| + } | |
| + // Make the lib directory as it exists after `make`. | |
| + if err := os.MkdirAll(filepath.Join(c.Dir, "lib"), 0755); err != nil { | |
| + return nil, err | |
| + } | |
| + } else if c.Args[0] == `make` && c.Args[1] == `archive` { | |
| + for _, arg := range c.Args { | |
| + if strings.HasPrefix(arg, `ARCHIVE=`) { | |
| + paths = append(paths, filepath.Join(c.Dir, strings.TrimPrefix(arg, `ARCHIVE=`))) | |
| + break | |
| } | |
| - paths = append(paths, filepath.Join(r.fakeBazelBin, "c-deps", "libgeos", libDir, lib+ext)) | |
| } | |
| } | |
| for _, path := range paths { | |
| - if err := os.MkdirAll(filepath.Dir(path), 0777); err != nil { | |
| - return nil, err | |
| - } | |
| if err := ioutil.WriteFile(path, []byte(cmd), 0666); err != nil { | |
| return nil, err | |
| } | |
| } | |
| + if len(paths) > 0 { | |
| + r.cmds = append(r.cmds, cmd) | |
| + } | |
| var output []byte | |
| return output, nil | |
| @@ -146,14 +130,10 @@ func TestProvisional(t *testing.T) { | |
| branch: `provisional_201901010101_v0.0.1-alpha`, | |
| }, | |
| expectedCmds: []string{ | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu official-binary v0.0.1-alpha release' -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| - "env=[MALLOC_CONF=prof:true] args=./cockroach.linux-2.6.32-gnu-amd64 version", | |
| - "env=[] args=ldd ./cockroach.linux-2.6.32-gnu-amd64", | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19 official-binary v0.0.1-alpha release' -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32 official-binary v0.0.1-alpha release' -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| + "env=[] args=[mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary BUILDINFO_TAG=v0.0.1-alpha BUILD_TAGGED_RELEASE=true]", | |
| + "env=[] args=[mkrelease darwin SUFFIX=.darwin-10.9-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary BUILDINFO_TAG=v0.0.1-alpha BUILD_TAGGED_RELEASE=true]", | |
| + "env=[] args=[mkrelease windows SUFFIX=.windows-6.2-amd64.exe GOFLAGS= TAGS= BUILDCHANNEL=official-binary BUILDINFO_TAG=v0.0.1-alpha BUILD_TAGGED_RELEASE=true]", | |
| + "env=[] args=[make archive ARCHIVE_BASE=cockroach-v0.0.1-alpha ARCHIVE=cockroach-v0.0.1-alpha.src.tgz BUILDINFO_TAG=v0.0.1-alpha]", | |
| }, | |
| expectedGets: nil, | |
| expectedPuts: []string{ | |
| @@ -163,6 +143,8 @@ func TestProvisional(t *testing.T) { | |
| "CONTENTS <binary stuff>", | |
| "s3://binaries.cockroachdb.com/cockroach-v0.0.1-alpha.windows-6.2-amd64.zip " + | |
| "CONTENTS <binary stuff>", | |
| + "s3://binaries.cockroachdb.com/cockroach-v0.0.1-alpha.src.tgz " + | |
| + "CONTENTS env=[] args=[make archive ARCHIVE_BASE=cockroach-v0.0.1-alpha ARCHIVE=cockroach-v0.0.1-alpha.src.tgz BUILDINFO_TAG=v0.0.1-alpha]", | |
| }, | |
| }, | |
| { | |
| @@ -174,40 +156,35 @@ func TestProvisional(t *testing.T) { | |
| sha: `00SHA00`, | |
| }, | |
| expectedCmds: []string{ | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu official-binary' -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| - "env=[MALLOC_CONF=prof:true] args=./cockroach.linux-2.6.32-gnu-amd64 version", | |
| - "env=[] args=ldd ./cockroach.linux-2.6.32-gnu-amd64", | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19 official-binary' -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| - "env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32 official-binary' -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| - "env=[] args=bazel info bazel-bin -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| + "env=[] args=[mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| + "env=[] args=[mkrelease darwin SUFFIX=.darwin-10.9-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| + "env=[] args=[mkrelease windows SUFFIX=.windows-6.2-amd64.exe GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| }, | |
| expectedGets: nil, | |
| expectedPuts: []string{ | |
| "s3://cockroach//cockroach/cockroach.linux-gnu-amd64.00SHA00 " + | |
| - "CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu official-binary' -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| + "CONTENTS env=[] args=[mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/cockroach.linux-gnu-amd64.LATEST/no-cache " + | |
| "REDIRECT /cockroach/cockroach.linux-gnu-amd64.00SHA00", | |
| - "s3://cockroach//cockroach/lib/libgeos.linux-gnu-amd64.00SHA00.so CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu official-binary' -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| + "s3://cockroach//cockroach/lib/libgeos.linux-gnu-amd64.00SHA00.so CONTENTS env=[] args=[mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos.linux-gnu-amd64.so.LATEST/no-cache REDIRECT /cockroach/lib/libgeos.linux-gnu-amd64.00SHA00.so", | |
| - "s3://cockroach//cockroach/lib/libgeos_c.linux-gnu-amd64.00SHA00.so CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-pc-linux-gnu official-binary' -c opt --config=ci --config=with_ui --config=crosslinuxbase", | |
| + "s3://cockroach//cockroach/lib/libgeos_c.linux-gnu-amd64.00SHA00.so CONTENTS env=[] args=[mkrelease linux-gnu SUFFIX=.linux-2.6.32-gnu-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos_c.linux-gnu-amd64.so.LATEST/no-cache REDIRECT /cockroach/lib/libgeos_c.linux-gnu-amd64.00SHA00.so", | |
| "s3://cockroach//cockroach/cockroach.darwin-amd64.00SHA00 " + | |
| - "CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19 official-binary' -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| + "CONTENTS env=[] args=[mkrelease darwin SUFFIX=.darwin-10.9-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/cockroach.darwin-amd64.LATEST/no-cache " + | |
| "REDIRECT /cockroach/cockroach.darwin-amd64.00SHA00", | |
| - "s3://cockroach//cockroach/lib/libgeos.darwin-amd64.00SHA00.dylib CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19 official-binary' -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| + "s3://cockroach//cockroach/lib/libgeos.darwin-amd64.00SHA00.dylib CONTENTS env=[] args=[mkrelease darwin SUFFIX=.darwin-10.9-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos.darwin-amd64.dylib.LATEST/no-cache REDIRECT /cockroach/lib/libgeos.darwin-amd64.00SHA00.dylib", | |
| - "s3://cockroach//cockroach/lib/libgeos_c.darwin-amd64.00SHA00.dylib CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-apple-darwin19 official-binary' -c opt --config=ci --config=with_ui --config=crossmacosbase", | |
| + "s3://cockroach//cockroach/lib/libgeos_c.darwin-amd64.00SHA00.dylib CONTENTS env=[] args=[mkrelease darwin SUFFIX=.darwin-10.9-amd64 GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos_c.darwin-amd64.dylib.LATEST/no-cache REDIRECT /cockroach/lib/libgeos_c.darwin-amd64.00SHA00.dylib", | |
| "s3://cockroach//cockroach/cockroach.windows-amd64.00SHA00.exe " + | |
| - "CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32 official-binary' -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| + "CONTENTS env=[] args=[mkrelease windows SUFFIX=.windows-6.2-amd64.exe GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/cockroach.windows-amd64.LATEST/no-cache " + | |
| "REDIRECT /cockroach/cockroach.windows-amd64.00SHA00.exe", | |
| - "s3://cockroach//cockroach/lib/libgeos.windows-amd64.00SHA00.dll CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32 official-binary' -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| + "s3://cockroach//cockroach/lib/libgeos.windows-amd64.00SHA00.dll CONTENTS env=[] args=[mkrelease windows SUFFIX=.windows-6.2-amd64.exe GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos.windows-amd64.dll.LATEST/no-cache REDIRECT /cockroach/lib/libgeos.windows-amd64.00SHA00.dll", | |
| - "s3://cockroach//cockroach/lib/libgeos_c.windows-amd64.00SHA00.dll CONTENTS env=[] args=bazel build //pkg/cmd/cockroach //c-deps:libgeos '--workspace_status_command=./build/bazelutil/stamp.sh x86_64-w64-mingw32 official-binary' -c opt --config=ci --config=with_ui --config=crosswindowsbase", | |
| + "s3://cockroach//cockroach/lib/libgeos_c.windows-amd64.00SHA00.dll CONTENTS env=[] args=[mkrelease windows SUFFIX=.windows-6.2-amd64.exe GOFLAGS= TAGS= BUILDCHANNEL=official-binary]", | |
| "s3://cockroach/cockroach/lib/libgeos_c.windows-amd64.dll.LATEST/no-cache REDIRECT /cockroach/lib/libgeos_c.windows-amd64.00SHA00.dll", | |
| }, | |
| }, | |
| @@ -219,13 +196,9 @@ func TestProvisional(t *testing.T) { | |
| var s3 mockS3 | |
| var exec mockExecRunner | |
| - fakeBazelBin, cleanup := testutils.TempDir(t) | |
| - defer cleanup() | |
| - exec.fakeBazelBin = fakeBazelBin | |
| flags := test.flags | |
| flags.pkgDir = dir | |
| - execFn := release.ExecFn{MockExecFn: exec.run} | |
| - run(&s3, flags, execFn) | |
| + run(&s3, exec.run, flags) | |
| require.Equal(t, test.expectedCmds, exec.cmds) | |
| require.Equal(t, test.expectedGets, s3.gets) | |
| require.Equal(t, test.expectedPuts, s3.puts) | |
| @@ -261,6 +234,7 @@ func TestBless(t *testing.T) { | |
| "s3://binaries.cockroachdb.com/cockroach-v0.0.1.linux-amd64.tgz", | |
| "s3://binaries.cockroachdb.com/cockroach-v0.0.1.darwin-10.9-amd64.tgz", | |
| "s3://binaries.cockroachdb.com/cockroach-v0.0.1.windows-6.2-amd64.zip", | |
| + "s3://binaries.cockroachdb.com/cockroach-v0.0.1.src.tgz", | |
| }, | |
| expectedPuts: []string{ | |
| "s3://binaries.cockroachdb.com/cockroach-latest.linux-amd64.tgz/no-cache " + | |
| @@ -269,6 +243,8 @@ func TestBless(t *testing.T) { | |
| "CONTENTS s3://binaries.cockroachdb.com/cockroach-v0.0.1.darwin-10.9-amd64.tgz", | |
| "s3://binaries.cockroachdb.com/cockroach-latest.windows-6.2-amd64.zip/no-cache " + | |
| "CONTENTS s3://binaries.cockroachdb.com/cockroach-v0.0.1.windows-6.2-amd64.zip", | |
| + "s3://binaries.cockroachdb.com/cockroach-latest.src.tgz/no-cache " + | |
| + "CONTENTS s3://binaries.cockroachdb.com/cockroach-v0.0.1.src.tgz", | |
| }, | |
| }, | |
| } | |
| @@ -277,7 +253,7 @@ func TestBless(t *testing.T) { | |
| t.Run(test.name, func(t *testing.T) { | |
| var s3 mockS3 | |
| var execFn release.ExecFn // bless shouldn't exec anything | |
| - run(&s3, test.flags, execFn) | |
| + run(&s3, execFn, test.flags) | |
| require.Equal(t, test.expectedGets, s3.gets) | |
| require.Equal(t, test.expectedPuts, s3.puts) | |
| }) | |
| diff --git a/pkg/cmd/release/BUILD.bazel b/pkg/cmd/release/BUILD.bazel | |
| index 54e721dfae..52fab9d313 100644 | |
| --- a/pkg/cmd/release/BUILD.bazel | |
| +++ b/pkg/cmd/release/BUILD.bazel | |
| @@ -1,4 +1,4 @@ | |
| -load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library", "go_test") | |
| +load("@io_bazel_rules_go//go:def.bzl", "go_binary", "go_library") | |
| go_library( | |
| name = "release_lib", | |
| @@ -6,11 +6,10 @@ go_library( | |
| "blockers.go", | |
| "git.go", | |
| "jira.go", | |
| + "mail.go", | |
| "main.go", | |
| "metadata.go", | |
| - "orchestration.go", | |
| "pick_sha.go", | |
| - "sender.go", | |
| "templates.go", | |
| ], | |
| importpath = "github.com/cockroachdb/cockroach/pkg/cmd/release", | |
| @@ -29,14 +28,3 @@ go_binary( | |
| embed = [":release_lib"], | |
| visibility = ["//visibility:public"], | |
| ) | |
| - | |
| -go_test( | |
| - name = "release_test", | |
| - srcs = ["sender_test.go"], | |
| - data = glob([ | |
| - "templates/**", | |
| - "testdata/**", | |
| - ]), | |
| - embed = [":release_lib"], | |
| - deps = ["@com_github_stretchr_testify//require"], | |
| -) | |
| diff --git a/pkg/cmd/release/blockers.go b/pkg/cmd/release/blockers.go | |
| index aa7069a1c4..8fc164f81f 100644 | |
| --- a/pkg/cmd/release/blockers.go | |
| +++ b/pkg/cmd/release/blockers.go | |
| @@ -13,34 +13,29 @@ package main | |
| import ( | |
| "fmt" | |
| "os" | |
| - "time" | |
| "github.com/spf13/cobra" | |
| ) | |
| const ( | |
| - // dateLayout is the expected date format for prepDate and publishDate input values. | |
| - dateLayout = "2006-01-02" | |
| - | |
| - // prepDate is the date when we expect to select the release candidate. | |
| - prepDate = "prep-date" | |
| - // publishDate is the date when we expect to publish the release candidate. | |
| - publishDate = "publish-date" | |
| - | |
| - // nextVersion can be left out for stable/patch releases, but needs to be passed in for pre-releases (alpha/beta/rc). | |
| - nextVersion = "next-version" | |
| + envSMTPUser = "SMTP_USER" | |
| + envSMTPPassword = "SMTP_PASSWORD" | |
| + envGithubToken = "GITHUB_TOKEN" | |
| + releaseSeries = "release-series" | |
| + smtpUser = "smtp-user" | |
| + smtpHost = "smtp-host" | |
| + smtpPort = "smtp-port" | |
| + emailAddresses = "to" | |
| + dryRun = "dry-run" | |
| ) | |
| var blockersFlags = struct { | |
| releaseSeries string | |
| - templatesDir string | |
| - prepDate string | |
| - publishDate string | |
| - nextVersion string | |
| smtpUser string | |
| smtpHost string | |
| smtpPort int | |
| emailAddresses []string | |
| + dryRun bool | |
| }{} | |
| var postReleaseSeriesBlockersCmd = &cobra.Command{ | |
| @@ -51,20 +46,16 @@ var postReleaseSeriesBlockersCmd = &cobra.Command{ | |
| } | |
| func init() { | |
| + // TODO: improve flag usage comments | |
| postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.releaseSeries, releaseSeries, "", "major release series") | |
| - postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.templatesDir, templatesDir, "", "templates directory") | |
| - postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.prepDate, prepDate, "", "date to select candidate") | |
| - postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.publishDate, publishDate, "", "date to publish candidate") | |
| - postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.nextVersion, nextVersion, "", "next release version") | |
| postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.smtpUser, smtpUser, os.Getenv(envSMTPUser), "SMTP user name") | |
| postReleaseSeriesBlockersCmd.Flags().StringVar(&blockersFlags.smtpHost, smtpHost, "", "SMTP host") | |
| postReleaseSeriesBlockersCmd.Flags().IntVar(&blockersFlags.smtpPort, smtpPort, 0, "SMTP port") | |
| postReleaseSeriesBlockersCmd.Flags().StringArrayVar(&blockersFlags.emailAddresses, emailAddresses, []string{}, "email addresses") | |
| + postReleaseSeriesBlockersCmd.Flags().BoolVar(&blockersFlags.dryRun, dryRun, false, "use dry run Jira project for issues") | |
| _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(releaseSeries) | |
| - _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(templatesDir) | |
| - _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(prepDate) | |
| - _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(publishDate) | |
| + _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(smtpUser) | |
| _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(smtpHost) | |
| _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(smtpPort) | |
| _ = postReleaseSeriesBlockersCmd.MarkFlagRequired(emailAddresses) | |
| @@ -79,70 +70,9 @@ func fetchReleaseSeriesBlockers(_ *cobra.Command, _ []string) error { | |
| if githubToken == "" { | |
| return fmt.Errorf("%s environment variable should be set", envGithubToken) | |
| } | |
| - if blockersFlags.smtpUser == "" { | |
| - return fmt.Errorf("either %s environment variable or %s flag should be set", envSMTPUser, smtpUser) | |
| - } | |
| - releasePrepDate, err := time.Parse(dateLayout, blockersFlags.prepDate) | |
| - if err != nil { | |
| - return fmt.Errorf("%s is not parseable into %s date layout", blockersFlags.prepDate, dateLayout) | |
| - } | |
| - releasePublishDate, err := time.Parse(dateLayout, blockersFlags.publishDate) | |
| - if err != nil { | |
| - return fmt.Errorf("%s is not parseable into %s date layout", blockersFlags.publishDate, dateLayout) | |
| - } | |
| - if blockersFlags.nextVersion == "" { | |
| - var err error | |
| - blockersFlags.nextVersion, err = findNextVersion(blockersFlags.releaseSeries) | |
| - if err != nil { | |
| - return fmt.Errorf("cannot find next release version: %w", err) | |
| - } | |
| - } | |
| - // TODO(celia): (future PR) fetch blockers and use real data instead of `temp*` values. | |
| - fmt.Println("TODO(celia): fetch blockers") | |
| - tempTotalBlockers := 23 | |
| - tempBlockerList := []ProjectBlocker{ | |
| - { | |
| - ProjectName: "Project ABC", | |
| - NumBlockers: 11, | |
| - }, | |
| - { | |
| - ProjectName: "Project XYZ", | |
| - NumBlockers: 12, | |
| - }, | |
| - } | |
| + // TODO(celia): fetch blockers and send out email | |
| + fmt.Println("TODO(celia): fetch blockers and send out email") | |
| - blockersURL := "go.crdb.dev/blockers/" + blockersFlags.releaseSeries | |
| - releaseBranch := "release-" + blockersFlags.releaseSeries | |
| - | |
| - // TODO(celia): (future PR) dynamically set branchExists, based on whether `releaseBranch` branch exists in crdb repo | |
| - branchExists := true | |
| - if !branchExists { | |
| - blockersURL = "go.crdb.dev/blockers" | |
| - releaseBranch = "master" | |
| - } | |
| - args := messageDataPostBlockers{ | |
| - Version: blockersFlags.nextVersion, | |
| - PrepDate: releasePrepDate.Format("Monday, January 2"), | |
| - ReleaseDate: releasePublishDate.Format("Monday, January 2"), | |
| - TotalBlockers: tempTotalBlockers, | |
| - BlockersURL: blockersURL, | |
| - ReleaseBranch: releaseBranch, | |
| - BlockerList: tempBlockerList, | |
| - } | |
| - opts := sendOpts{ | |
| - templatesDir: blockersFlags.templatesDir, | |
| - from: fmt.Sprintf("Justin Beaver <%s>", blockersFlags.smtpUser), | |
| - host: blockersFlags.smtpHost, | |
| - port: blockersFlags.smtpPort, | |
| - user: blockersFlags.smtpUser, | |
| - password: smtpPassword, | |
| - to: blockersFlags.emailAddresses, | |
| - } | |
| - | |
| - fmt.Println("Sending email") | |
| - if err := sendMailPostBlockers(args, opts); err != nil { | |
| - return fmt.Errorf("cannot send email: %w", err) | |
| - } | |
| return nil | |
| } | |
| diff --git a/pkg/cmd/release/git.go b/pkg/cmd/release/git.go | |
| index 869e87cd53..778b531c99 100644 | |
| --- a/pkg/cmd/release/git.go | |
| +++ b/pkg/cmd/release/git.go | |
| @@ -31,19 +31,6 @@ type releaseInfo struct { | |
| releaseSeries string | |
| } | |
| -// findNextVersion returns the next release version for given releaseSeries. | |
| -func findNextVersion(releaseSeries string) (string, error) { | |
| - prevReleaseVersion, err := findPreviousRelease(releaseSeries) | |
| - if err != nil { | |
| - return "", fmt.Errorf("cannot find previous release: %w", err) | |
| - } | |
| - nextReleaseVersion, err := bumpVersion(prevReleaseVersion) | |
| - if err != nil { | |
| - return "", fmt.Errorf("cannot bump version: %w", err) | |
| - } | |
| - return nextReleaseVersion, nil | |
| -} | |
| - | |
| // findNextRelease finds all required information for the next release. | |
| func findNextRelease(releaseSeries string) (releaseInfo, error) { | |
| prevReleaseVersion, err := findPreviousRelease(releaseSeries) | |
| diff --git a/pkg/cmd/release/mail.go b/pkg/cmd/release/mail.go | |
| new file mode 100644 | |
| index 0000000000..fb632ba200 | |
| --- /dev/null | |
| +++ b/pkg/cmd/release/mail.go | |
| @@ -0,0 +1,94 @@ | |
| +// Copyright 2019 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package main | |
| + | |
| +import ( | |
| + "fmt" | |
| + htmltemplate "html/template" | |
| + "net/smtp" | |
| + "net/textproto" | |
| + | |
| + "github.com/jordan-wright/email" | |
| +) | |
| + | |
| +var emailSubjectTemplate = "Release {{ .Version }}" | |
| +var emailTextTemplate = ` | |
| +A candidate SHA has been selected for {{ .Version }}. Proceeding to qualification shortly. | |
| + | |
| + SHA: {{ .SHA }} | |
| + Tracking Issue: {{ .TrackingIssueURL }} | |
| + List of changes since last release: {{ .DiffURL }} | |
| + | |
| +Thanks | |
| +Release Engineering | |
| +` | |
| +var emailHTMLTemplate = ` | |
| +<html> | |
| +<body> | |
| +<p>A candidate SHA has been selected for <strong>{{ .Version }}</strong>. Proceeding to qualification shortly.</p> | |
| +<ul> | |
| +<li>SHA: <a href="https://github.com/cockroachlabs/release-staging/commit/{{ .SHA }}">{{ .SHA }}</a></li> | |
| +<li>Tracking Issue: <a href="{{ .TrackingIssueURL }}">{{ .TrackingIssue }}</a></li> | |
| +<li><a href="{{ .DiffURL }}">List of changes</a> since last release</li> | |
| +</ul> | |
| +<p>Thanks,<br /> | |
| +Release Engineering</p> | |
| +</body> | |
| +</html> | |
| +` | |
| + | |
| +type emailArgs struct { | |
| + Version string | |
| + SHA string | |
| + TrackingIssue string | |
| + TrackingIssueURL htmltemplate.URL | |
| + DiffURL htmltemplate.URL | |
| +} | |
| + | |
| +type smtpOpts struct { | |
| + host string | |
| + port int | |
| + user string | |
| + password string | |
| + from string | |
| + to []string | |
| +} | |
| + | |
| +// sendmail creates and sends an email to the releases mailing list | |
| +func sendmail(args emailArgs, smtpOpts smtpOpts) error { | |
| + text, err := templateToText(emailTextTemplate, args) | |
| + if err != nil { | |
| + return fmt.Errorf("cannot use text template: %w", err) | |
| + } | |
| + subject, err := templateToText(emailSubjectTemplate, args) | |
| + if err != nil { | |
| + return fmt.Errorf("cannot use subject template: %w", err) | |
| + } | |
| + html, err := templateToHTML(emailHTMLTemplate, args) | |
| + if err != nil { | |
| + return fmt.Errorf("cannot use html template: %w", err) | |
| + } | |
| + | |
| + e := &email.Email{ | |
| + To: smtpOpts.to, | |
| + From: smtpOpts.from, | |
| + Subject: subject, | |
| + Text: []byte(text), | |
| + HTML: []byte(html), | |
| + Headers: textproto.MIMEHeader{}, | |
| + } | |
| + emailAuth := smtp.PlainAuth("", smtpOpts.user, smtpOpts.password, smtpOpts.host) | |
| + addr := fmt.Sprintf("%s:%d", smtpOpts.host, smtpOpts.port) | |
| + if err := e.Send(addr, emailAuth); err != nil { | |
| + return fmt.Errorf("cannot send email: %w", err) | |
| + } | |
| + return nil | |
| +} | |
| diff --git a/pkg/cmd/release/main.go b/pkg/cmd/release/main.go | |
| index 69c67f02ba..be2e193f3a 100644 | |
| --- a/pkg/cmd/release/main.go | |
| +++ b/pkg/cmd/release/main.go | |
| @@ -14,19 +14,6 @@ import "github.com/spf13/cobra" | |
| var rootCmd = &cobra.Command{Use: "release"} | |
| -const ( | |
| - envSMTPUser = "SMTP_USER" | |
| - envSMTPPassword = "SMTP_PASSWORD" | |
| - envGithubToken = "GITHUB_TOKEN" | |
| - releaseSeries = "release-series" | |
| - templatesDir = "template-dir" | |
| - smtpUser = "smtp-user" | |
| - smtpHost = "smtp-host" | |
| - smtpPort = "smtp-port" | |
| - emailAddresses = "to" | |
| - dryRun = "dry-run" | |
| -) | |
| - | |
| func main() { | |
| if err := rootCmd.Execute(); err != nil { | |
| panic(err) | |
| @@ -36,5 +23,4 @@ func main() { | |
| func init() { | |
| rootCmd.AddCommand(pickSHACmd) | |
| rootCmd.AddCommand(postReleaseSeriesBlockersCmd) | |
| - rootCmd.AddCommand(setOrchestrationVersionCmd) | |
| } | |
| diff --git a/pkg/cmd/release/orchestration.go b/pkg/cmd/release/orchestration.go | |
| deleted file mode 100644 | |
| index 0dfc901c81..0000000000 | |
| --- a/pkg/cmd/release/orchestration.go | |
| +++ /dev/null | |
| @@ -1,100 +0,0 @@ | |
| -// Copyright 2019 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package main | |
| - | |
| -import ( | |
| - "errors" | |
| - "fmt" | |
| - "io/ioutil" | |
| - "os" | |
| - "path" | |
| - "path/filepath" | |
| - "strings" | |
| - | |
| - "github.com/spf13/cobra" | |
| -) | |
| - | |
| -const ( | |
| - versionFlag = "version" | |
| - templatesDirFlag = "template-dir" | |
| - outputDirFlag = "output-dir" | |
| -) | |
| - | |
| -var orchestrationFlags = struct { | |
| - version string | |
| - templatesDir string | |
| - outputDir string | |
| -}{} | |
| - | |
| -var setOrchestrationVersionCmd = &cobra.Command{ | |
| - Use: "set-orchestration-version", | |
| - Short: "Set orchestration version", | |
| - Long: "Updates orchestration version under the ./cloud/kubernetes directory", | |
| - RunE: setOrchestrationVersion, | |
| -} | |
| - | |
| -func init() { | |
| - setOrchestrationVersionCmd.Flags().StringVar(&orchestrationFlags.version, versionFlag, "", "cockroachdb version") | |
| - setOrchestrationVersionCmd.Flags().StringVar(&orchestrationFlags.templatesDir, templatesDirFlag, "", | |
| - "orchestration templates directory") | |
| - setOrchestrationVersionCmd.Flags().StringVar(&orchestrationFlags.outputDir, outputDirFlag, "", | |
| - "orchestration directory") | |
| - _ = setOrchestrationVersionCmd.MarkFlagRequired(versionFlag) | |
| - _ = setOrchestrationVersionCmd.MarkFlagRequired(templatesDirFlag) | |
| - _ = setOrchestrationVersionCmd.MarkFlagRequired(outputDirFlag) | |
| -} | |
| - | |
| -func setOrchestrationVersion(_ *cobra.Command, _ []string) error { | |
| - dirInfo, err := os.Stat(orchestrationFlags.templatesDir) | |
| - if err != nil { | |
| - return fmt.Errorf("cannot stat templates directory: %w", err) | |
| - } | |
| - if !dirInfo.IsDir() { | |
| - return fmt.Errorf("%s is not a directory", orchestrationFlags.templatesDir) | |
| - } | |
| - return filepath.Walk(orchestrationFlags.templatesDir, func(filePath string, fileInfo os.FileInfo, e error) error { | |
| - if e != nil { | |
| - return e | |
| - } | |
| - // Skip directories | |
| - if !fileInfo.Mode().IsRegular() { | |
| - return nil | |
| - } | |
| - // calculate file directory relative to the given root directory. | |
| - dir := path.Dir(filePath) | |
| - relDir, err := filepath.Rel(orchestrationFlags.templatesDir, dir) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - destDir := filepath.Join(orchestrationFlags.outputDir, relDir) | |
| - destFile := filepath.Join(destDir, fileInfo.Name()) | |
| - if err := os.MkdirAll(destDir, 0755); err != nil && !errors.Is(err, os.ErrExist) { | |
| - return err | |
| - } | |
| - contents, err := ioutil.ReadFile(filePath) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - // Go templates cannot be used here, because some files are templates already. | |
| - generatedContents := strings.ReplaceAll(string(contents), "@VERSION@", orchestrationFlags.version) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if strings.HasSuffix(destFile, ".yaml") { | |
| - generatedContents = fmt.Sprintf("# Generated file, DO NOT EDIT. Source: %s\n", filePath) + generatedContents | |
| - } | |
| - err = ioutil.WriteFile(destFile, []byte(generatedContents), fileInfo.Mode().Perm()) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - return nil | |
| - }) | |
| -} | |
| diff --git a/pkg/cmd/release/pick_sha.go b/pkg/cmd/release/pick_sha.go | |
| index 4cd48e166b..620f27d3c8 100644 | |
| --- a/pkg/cmd/release/pick_sha.go | |
| +++ b/pkg/cmd/release/pick_sha.go | |
| @@ -19,20 +19,12 @@ import ( | |
| "github.com/spf13/cobra" | |
| ) | |
| -const ( | |
| - qualifyBucket = "qualify-bucket" | |
| - qualifyObjectPrefix = "qualify-object-prefix" | |
| - releaseBucket = "release-bucket" | |
| - releaseObjectPrefix = "release-object-prefix" | |
| -) | |
| - | |
| var pickSHAFlags = struct { | |
| qualifyBucket string | |
| qualifyObjectPrefix string | |
| releaseBucket string | |
| releaseObjectPrefix string | |
| releaseSeries string | |
| - templatesDir string | |
| smtpUser string | |
| smtpHost string | |
| smtpPort int | |
| @@ -50,29 +42,27 @@ var pickSHACmd = &cobra.Command{ | |
| func init() { | |
| // TODO: improve flag usage comments | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.qualifyBucket, qualifyBucket, "", "release qualification metadata GCS bucket") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.qualifyObjectPrefix, qualifyObjectPrefix, "", | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.qualifyBucket, "qualify-bucket", "", "release qualification metadata GCS bucket") | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.qualifyObjectPrefix, "qualify-object-prefix", "", | |
| "release qualification object prefix") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseBucket, releaseBucket, "", "release candidates metadata GCS bucket") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseObjectPrefix, releaseObjectPrefix, "", "release candidate object prefix") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseSeries, releaseSeries, "", "major release series") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.templatesDir, templatesDir, "", "templates directory") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.smtpUser, smtpUser, os.Getenv(envSMTPUser), "SMTP user name") | |
| - pickSHACmd.Flags().StringVar(&pickSHAFlags.smtpHost, smtpHost, "", "SMTP host") | |
| - pickSHACmd.Flags().IntVar(&pickSHAFlags.smtpPort, smtpPort, 0, "SMTP port") | |
| - pickSHACmd.Flags().StringArrayVar(&pickSHAFlags.emailAddresses, emailAddresses, []string{}, "email addresses") | |
| - pickSHACmd.Flags().BoolVar(&pickSHAFlags.dryRun, dryRun, false, "use dry run Jira project for issues") | |
| - | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseBucket, "release-bucket", "", "release candidates metadata GCS bucket") | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseObjectPrefix, "release-object-prefix", "", "release candidate object prefix") | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.releaseSeries, "release-series", "", "major release series") | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.smtpUser, "smtp-user", os.Getenv("SMTP_USER"), "SMTP user name") | |
| + pickSHACmd.Flags().StringVar(&pickSHAFlags.smtpHost, "smtp-host", "", "SMTP host") | |
| + pickSHACmd.Flags().IntVar(&pickSHAFlags.smtpPort, "smtp-port", 0, "SMTP port") | |
| + pickSHACmd.Flags().StringArrayVar(&pickSHAFlags.emailAddresses, "to", []string{}, "email addresses") | |
| + pickSHACmd.Flags().BoolVar(&pickSHAFlags.dryRun, "dry-run", false, "use dry run Jira project for issues") | |
| requiredFlags := []string{ | |
| - qualifyBucket, | |
| - qualifyObjectPrefix, | |
| - releaseBucket, | |
| - releaseObjectPrefix, | |
| - releaseSeries, | |
| - smtpUser, | |
| - smtpHost, | |
| - smtpPort, | |
| - emailAddresses, | |
| + "qualify-bucket", | |
| + "qualify-object-prefix", | |
| + "release-bucket", | |
| + "release-object-prefix", | |
| + "release-series", | |
| + "smtp-user", | |
| + "smtp-host", | |
| + "smtp-port", | |
| + "to", | |
| } | |
| for _, flag := range requiredFlags { | |
| if err := pickSHACmd.MarkFlagRequired(flag); err != nil { | |
| @@ -130,24 +120,23 @@ func pickSHA(_ *cobra.Command, _ []string) error { | |
| fmt.Sprintf("https://github.com/cockroachdb/cockroach/compare/%s...%s", | |
| nextRelease.prevReleaseVersion, | |
| nextRelease.buildInfo.SHA)) | |
| - args := messageDataPickSHA{ | |
| + args := emailArgs{ | |
| Version: nextRelease.nextReleaseVersion, | |
| SHA: nextRelease.buildInfo.SHA, | |
| TrackingIssue: trackingIssue.Key, | |
| TrackingIssueURL: template.URL(trackingIssue.url()), | |
| DiffURL: diffURL, | |
| } | |
| - opts := sendOpts{ | |
| - templatesDir: pickSHAFlags.templatesDir, | |
| - from: fmt.Sprintf("Justin Beaver <%s>", pickSHAFlags.smtpUser), | |
| - host: pickSHAFlags.smtpHost, | |
| - port: pickSHAFlags.smtpPort, | |
| - user: pickSHAFlags.smtpUser, | |
| - password: smtpPassword, | |
| - to: pickSHAFlags.emailAddresses, | |
| + opts := smtpOpts{ | |
| + from: fmt.Sprintf("Justin Beaver <%s>", pickSHAFlags.smtpUser), | |
| + host: pickSHAFlags.smtpHost, | |
| + port: pickSHAFlags.smtpPort, | |
| + user: pickSHAFlags.smtpUser, | |
| + password: smtpPassword, | |
| + to: pickSHAFlags.emailAddresses, | |
| } | |
| fmt.Println("Sending email") | |
| - if err := sendMailPickSHA(args, opts); err != nil { | |
| + if err := sendmail(args, opts); err != nil { | |
| return fmt.Errorf("cannot send email: %w", err) | |
| } | |
| return nil | |
| diff --git a/pkg/cmd/release/sender.go b/pkg/cmd/release/sender.go | |
| deleted file mode 100644 | |
| index b090f86b94..0000000000 | |
| --- a/pkg/cmd/release/sender.go | |
| +++ /dev/null | |
| @@ -1,189 +0,0 @@ | |
| -// Copyright 2019 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package main | |
| - | |
| -import ( | |
| - "fmt" | |
| - htmltemplate "html/template" | |
| - "net/smtp" | |
| - "net/textproto" | |
| - "os" | |
| - "path/filepath" | |
| - "strings" | |
| - | |
| - "github.com/jordan-wright/email" | |
| -) | |
| - | |
| -const ( | |
| - templatePrefixPickSHA = "pick-sha" | |
| - templatePrefixPostBlockers = "post-blockers" | |
| - templatePrefixPostBlockersAlpha = "post-blockers.alpha" | |
| -) | |
| - | |
| -type messageDataPickSHA struct { | |
| - Version string | |
| - SHA string | |
| - TrackingIssue string | |
| - TrackingIssueURL htmltemplate.URL | |
| - DiffURL htmltemplate.URL | |
| -} | |
| - | |
| -// ProjectBlocker lists the number of blockers per project. | |
| -// This needs to be public in order to be used by the html/template engine. | |
| -type ProjectBlocker struct { | |
| - ProjectName string | |
| - NumBlockers int | |
| -} | |
| - | |
| -type messageDataPostBlockers struct { | |
| - Version string | |
| - PrepDate string | |
| - ReleaseDate string | |
| - TotalBlockers int | |
| - BlockersURL string | |
| - ReleaseBranch string | |
| - BlockerList []ProjectBlocker | |
| -} | |
| - | |
| -type postBlockerTemplateArgs struct { | |
| - BackportsUseBackboard bool | |
| - BackportsWeeklyTriageReview bool | |
| -} | |
| - | |
| -type message struct { | |
| - Subject string | |
| - TextBody string | |
| - HTMLBody string | |
| -} | |
| - | |
| -func loadTemplate(templatesDir, template string) (string, error) { | |
| - file, err := os.ReadFile(filepath.Join(templatesDir, template)) | |
| - if err != nil { | |
| - return "", fmt.Errorf("loadTemplate %s: %w", template, err) | |
| - } | |
| - return string(file), nil | |
| -} | |
| - | |
| -// newMessage generates new message parts, based on: | |
| -// - templatePrefix - the filename prefix for subject/txt/html templates in the ./templates/ folder | |
| -// - data - the data object applied to the html/text/subject templates | |
| -func newMessage(templatesDir string, templatePrefix string, data interface{}) (*message, error) { | |
| - subjectTemplate, err := loadTemplate(templatesDir, templatePrefix+".subject") | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - subject, err := templateToText(subjectTemplate, data) | |
| - if err != nil { | |
| - return nil, fmt.Errorf("templateToText %s: %w", templatePrefix+".subject", err) | |
| - } | |
| - | |
| - textTemplate, err := loadTemplate(templatesDir, templatePrefix+".txt") | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - text, err := templateToText(textTemplate, data) | |
| - if err != nil { | |
| - return nil, fmt.Errorf("templateToText %s: %w", templatePrefix+".txt", err) | |
| - } | |
| - | |
| - htmlTemplate, err := loadTemplate(templatesDir, templatePrefix+".gohtml") | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - html, err := templateToHTML(htmlTemplate, data) | |
| - if err != nil { | |
| - return nil, fmt.Errorf("templateToHTML %s: %w", templatePrefix+".gohtml", err) | |
| - } | |
| - | |
| - return &message{ | |
| - Subject: subject, | |
| - TextBody: text, | |
| - HTMLBody: html, | |
| - }, nil | |
| -} | |
| - | |
| -type sendOpts struct { | |
| - templatesDir string | |
| - host string | |
| - port int | |
| - user string | |
| - password string | |
| - from string | |
| - to []string | |
| -} | |
| - | |
| -func sendMailPostBlockers(args messageDataPostBlockers, opts sendOpts) error { | |
| - templatePrefix := templatePrefixPostBlockers | |
| - | |
| - // Backport policy: | |
| - // - stable/production: refer to backboard | |
| - // - alpha: no need to mention backports | |
| - // - beta/rc's: backports reviewed during triage meeting | |
| - backportsUseBackboard := false | |
| - backportsWeeklyTriageReview := false | |
| - | |
| - switch { | |
| - case strings.Contains(args.Version, "-alpha."): | |
| - // alpha copy is so different that we'll use a completely separate template. | |
| - templatePrefix = templatePrefixPostBlockersAlpha | |
| - case | |
| - strings.Contains(args.Version, "-beta."), | |
| - strings.Contains(args.Version, "-rc."): | |
| - backportsWeeklyTriageReview = true | |
| - default: // stable/production | |
| - backportsUseBackboard = true | |
| - } | |
| - | |
| - data := struct { | |
| - Args messageDataPostBlockers | |
| - Template postBlockerTemplateArgs | |
| - }{ | |
| - Args: args, | |
| - Template: postBlockerTemplateArgs{ | |
| - BackportsUseBackboard: backportsUseBackboard, | |
| - BackportsWeeklyTriageReview: backportsWeeklyTriageReview, | |
| - }, | |
| - } | |
| - msg, err := newMessage(opts.templatesDir, templatePrefix, data) | |
| - if err != nil { | |
| - return fmt.Errorf("newMessage: %w", err) | |
| - } | |
| - return sendmail(msg, opts) | |
| -} | |
| - | |
| -func sendMailPickSHA(args messageDataPickSHA, opts sendOpts) error { | |
| - msg, err := newMessage(opts.templatesDir, templatePrefixPickSHA, args) | |
| - if err != nil { | |
| - return fmt.Errorf("newMessage: %w", err) | |
| - } | |
| - return sendmail(msg, opts) | |
| -} | |
| - | |
| -// sendmail creates and sends an email to the releases mailing list. | |
| -// sendmail is specified as a function closure to allow for testing | |
| -// of sendMail* methods. | |
| -var sendmail = func(content *message, smtpOpts sendOpts) error { | |
| - | |
| - e := &email.Email{ | |
| - To: smtpOpts.to, | |
| - From: smtpOpts.from, | |
| - Subject: content.Subject, | |
| - Text: []byte(content.TextBody), | |
| - HTML: []byte(content.HTMLBody), | |
| - Headers: textproto.MIMEHeader{}, | |
| - } | |
| - emailAuth := smtp.PlainAuth("", smtpOpts.user, smtpOpts.password, smtpOpts.host) | |
| - addr := fmt.Sprintf("%s:%d", smtpOpts.host, smtpOpts.port) | |
| - if err := e.Send(addr, emailAuth); err != nil { | |
| - return fmt.Errorf("cannot send email: %w", err) | |
| - } | |
| - return nil | |
| -} | |
| diff --git a/pkg/cmd/release/sender_test.go b/pkg/cmd/release/sender_test.go | |
| deleted file mode 100644 | |
| index 931cbf93bc..0000000000 | |
| --- a/pkg/cmd/release/sender_test.go | |
| +++ /dev/null | |
| @@ -1,259 +0,0 @@ | |
| -// Copyright 2019 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package main | |
| - | |
| -import ( | |
| - "flag" | |
| - "fmt" | |
| - htmltemplate "html/template" | |
| - "io/ioutil" | |
| - "path/filepath" | |
| - "reflect" | |
| - "testing" | |
| - | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -var update = flag.Bool("update", false, "update .golden files") | |
| - | |
| -// hookGlobal sets `*ptr = val` and returns a closure for restoring `*ptr` to | |
| -// its original value. A runtime panic will occur if `val` is not assignable to | |
| -// `*ptr`. | |
| -func hookGlobal(ptr, val interface{}) func() { | |
| - global := reflect.ValueOf(ptr).Elem() | |
| - orig := reflect.New(global.Type()).Elem() | |
| - orig.Set(global) | |
| - global.Set(reflect.ValueOf(val)) | |
| - return func() { global.Set(orig) } | |
| -} | |
| - | |
| -func TestPickSHA(t *testing.T) { | |
| - var expectedMessage *message | |
| - defer hookGlobal( | |
| - &sendmail, | |
| - func(content *message, smtpOpts sendOpts) error { | |
| - expectedMessage = content | |
| - return nil | |
| - })() | |
| - | |
| - args := messageDataPickSHA{ | |
| - Version: "v21.1.13", | |
| - SHA: "0fd6eead6c6eb7b2529deb39197cc3c95e93ded8", | |
| - TrackingIssue: "REL-111", | |
| - TrackingIssueURL: htmltemplate.URL("https://cockroachlabs.atlassian.net/browse/REL-111"), | |
| - DiffURL: "https://github.com/cockroachdb/cockroach/compare/v21.1.13...0fd6eead6c6eb7b2529deb39197cc3c95e93ded8", | |
| - } | |
| - require.NoError(t, sendMailPickSHA( | |
| - args, sendOpts{templatesDir: "templates"}, | |
| - )) | |
| - verifyMessageValues(t, expectedMessage, templatePrefixPickSHA) | |
| -} | |
| - | |
| -func TestPostBlockers(t *testing.T) { | |
| - tests := []struct { | |
| - testCase string | |
| - version string | |
| - blockersPerProject []int | |
| - goldenFilePrefix string | |
| - expectedContains []string | |
| - expectedNotContains []string | |
| - }{ | |
| - { | |
| - testCase: "alpha: zero-blockers", | |
| - version: "v19.1.0-alpha.3", | |
| - blockersPerProject: []int{}, | |
| - goldenFilePrefix: templatePrefixPostBlockersAlpha + ".zero-blockers", | |
| - expectedContains: []string{ | |
| - "We are clear to proceed with preparation and qualification", | |
| - }, | |
| - }, | |
| - { | |
| - testCase: "alpha: 1-blocker", | |
| - version: "v19.1.0-alpha.3", | |
| - blockersPerProject: []int{1}, | |
| - goldenFilePrefix: templatePrefixPostBlockersAlpha + ".1-blocker", | |
| - expectedNotContains: []string{ | |
| - "which must be resolved before a candidate is chosen", | |
| - }, | |
| - }, | |
| - { | |
| - testCase: "alpha: many-blockers", | |
| - version: "v19.1.0-alpha.3", | |
| - blockersPerProject: []int{4, 2, 3}, | |
| - goldenFilePrefix: templatePrefixPostBlockersAlpha + ".many-blockers", | |
| - expectedNotContains: []string{ | |
| - "which must be resolved before a candidate is chosen", | |
| - "reminder to merge any outstanding backports", | |
| - "reviewed at Monday's triage meeting", | |
| - }, | |
| - }, | |
| - { | |
| - testCase: "non-alpha: zero-blockers. stable/production: refer to backboard", | |
| - version: "v19.1.11", | |
| - blockersPerProject: []int{}, | |
| - goldenFilePrefix: templatePrefixPostBlockers + ".zero-blockers", | |
| - expectedContains: []string{ | |
| - "We are clear to proceed with preparation and qualification", | |
| - "backboard.crdb.dev", | |
| - }, | |
| - expectedNotContains: []string{ | |
| - "reviewed at Monday's triage meeting", | |
| - }, | |
| - }, | |
| - { | |
| - testCase: "non-alpha: 1-blocker. beta/rc's are reviewed in triage meeting", | |
| - version: "v19.1.0-beta.2", | |
| - blockersPerProject: []int{1}, | |
| - goldenFilePrefix: templatePrefixPostBlockers + ".1-blocker", | |
| - expectedContains: []string{ | |
| - "which must be resolved before a candidate is chosen", | |
| - "reviewed at Monday's triage meeting", | |
| - }, | |
| - expectedNotContains: []string{ | |
| - "backboard.crdb.dev", | |
| - }, | |
| - }, | |
| - { | |
| - testCase: "non-alpha: many-blockers. beta/rc's are reviewed in triage meeting", | |
| - version: "v19.1.0-rc.3", | |
| - blockersPerProject: []int{4, 2, 3}, | |
| - goldenFilePrefix: templatePrefixPostBlockers + ".many-blockers", | |
| - expectedContains: []string{ | |
| - "which must be resolved before a candidate is chosen", | |
| - "reviewed at Monday's triage meeting", | |
| - }, | |
| - expectedNotContains: []string{ | |
| - "backboard.crdb.dev", | |
| - }, | |
| - }, | |
| - } | |
| - | |
| - for _, test := range tests { | |
| - t.Run(fmt.Sprintf(test.testCase), func(t *testing.T) { | |
| - var expectedMessage *message | |
| - defer hookGlobal( | |
| - &sendmail, | |
| - func(content *message, smtpOpts sendOpts) error { | |
| - expectedMessage = content | |
| - return nil | |
| - })() | |
| - | |
| - // generate test data for blockerList | |
| - totalBlockers := 0 | |
| - var blockerList []ProjectBlocker | |
| - for i, numBlockers := range test.blockersPerProject { | |
| - blockerList = append(blockerList, ProjectBlocker{ | |
| - ProjectName: fmt.Sprintf("Project %d", i+1), | |
| - NumBlockers: numBlockers, | |
| - }) | |
| - totalBlockers += numBlockers | |
| - } | |
| - args := messageDataPostBlockers{ | |
| - Version: test.version, | |
| - PrepDate: "Saturday, April 1", | |
| - ReleaseDate: "Saturday, April 11", | |
| - TotalBlockers: totalBlockers, | |
| - BlockersURL: "go.crdb.dev/blockers", | |
| - ReleaseBranch: "master", | |
| - BlockerList: blockerList, | |
| - } | |
| - require.NoError(t, sendMailPostBlockers( | |
| - args, sendOpts{templatesDir: "templates"}, | |
| - )) | |
| - verifyMessageValues(t, expectedMessage, test.goldenFilePrefix) | |
| - for _, expectedContains := range test.expectedContains { | |
| - require.Contains(t, expectedMessage.HTMLBody, expectedContains) | |
| - require.Contains(t, expectedMessage.TextBody, expectedContains) | |
| - } | |
| - for _, expectedNotContains := range test.expectedNotContains { | |
| - require.NotContains(t, expectedMessage.HTMLBody, expectedNotContains) | |
| - require.NotContains(t, expectedMessage.TextBody, expectedNotContains) | |
| - } | |
| - }) | |
| - } | |
| -} | |
| - | |
| -func verifyMessageValue(t *testing.T, expected string, goldenFileName string) { | |
| - templates := mockMailTemplates() | |
| - | |
| - if *update { | |
| - templates.CreateGolden(t, expected, goldenFileName) | |
| - } | |
| - | |
| - goldenFile := templates.GetGolden(t, goldenFileName) | |
| - | |
| - require.Equal(t, goldenFile, expected) | |
| -} | |
| - | |
| -func verifyMessageValues(t *testing.T, msg *message, goldenFilePrefix string) { | |
| - goldenTextFileName := goldenFilePrefix + textBodyGoldenFileSuffix | |
| - goldenHTMLFileName := goldenFilePrefix + htmlBodyGoldenFileSuffix | |
| - goldenSubjectFileName := goldenFilePrefix + subjectGoldenFileSuffix | |
| - | |
| - verifyMessageValue(t, msg.TextBody, goldenTextFileName) | |
| - verifyMessageValue(t, msg.HTMLBody, goldenHTMLFileName) | |
| - verifyMessageValue(t, msg.Subject, goldenSubjectFileName) | |
| -} | |
| - | |
| -type mailTemplates struct { | |
| -} | |
| - | |
| -func mockMailTemplates() *mailTemplates { | |
| - return &mailTemplates{} | |
| -} | |
| - | |
| -const ( | |
| - subjectGoldenFileSuffix = ".subject.golden" | |
| - textBodyGoldenFileSuffix = ".txt.golden" | |
| - htmlBodyGoldenFileSuffix = ".html.golden" | |
| - testDataDirectory = "testdata" | |
| -) | |
| - | |
| -/* | |
| - | |
| -Golden Files | |
| - | |
| -Golden files are files containing expected output (like Snapshots in Jest). We use them here to | |
| -compare expected output of email templates to generated output. For every email that is sent there | |
| -are three associated files; an HTML file, a Plain Text file, and a file containing the Subject. | |
| -Testing these files can be done with the convinience method `verifyMessageValues`. For Example, | |
| - | |
| - func blahblahEmailTest (t *testing.T) { | |
| - ... | |
| - | |
| - messageValues := MessageValues{ | |
| - Subject: "blah blah subject", | |
| - TextBody: "blah blah text body", | |
| - HtmlBody: "<main>blah blah html stuff</main>", | |
| - } | |
| - goldenFilePrefix := "very_cool_email" | |
| - | |
| - verifyMessageValues(t, MessageValues, goldenFilePrefix) | |
| - } | |
| - | |
| -To generate golden files you may run go test with the update flag | |
| - | |
| - go test -v ./pkg/cmd/release -update | |
| - | |
| -*/ | |
| - | |
| -// CreateGolden generates Jest-like snapshots used to verify Mailer message formatting. | |
| -func (mt *mailTemplates) CreateGolden(t *testing.T, actualValue string, filename string) { | |
| - err := ioutil.WriteFile(filepath.Join(testDataDirectory, filename), []byte(actualValue), 0644) | |
| - require.NoError(t, err) | |
| -} | |
| - | |
| -// GetGolden verifies Mailer-generated messages against golden files generated by CreateGolden. | |
| -func (mt *mailTemplates) GetGolden(t *testing.T, filename string) string { | |
| - file, err := ioutil.ReadFile(filepath.Join(testDataDirectory, filename)) | |
| - require.NoError(t, err) | |
| - return string(file) | |
| -} | |
| diff --git a/pkg/cmd/release/templates/pick-sha.gohtml b/pkg/cmd/release/templates/pick-sha.gohtml | |
| deleted file mode 100644 | |
| index e12b4bacbd..0000000000 | |
| --- a/pkg/cmd/release/templates/pick-sha.gohtml | |
| +++ /dev/null | |
| @@ -1,14 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>A candidate SHA has been selected for <strong>{{ .Version }}</strong>. Proceeding to qualification shortly.</p> | |
| - <ul> | |
| - <li>SHA: <a href="https://github.com/cockroachlabs/release-staging/commit/{{ .SHA }}">{{ .SHA }}</a></li> | |
| - <li>Tracking Issue: <a href="{{ .TrackingIssueURL }}">{{ .TrackingIssue }}</a></li> | |
| - <li><a href="{{ .DiffURL }}">List of changes</a> since last release</li> | |
| - </ul> | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/templates/pick-sha.subject b/pkg/cmd/release/templates/pick-sha.subject | |
| deleted file mode 100644 | |
| index 27e01766e8..0000000000 | |
| --- a/pkg/cmd/release/templates/pick-sha.subject | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release {{ .Version }} | |
| diff --git a/pkg/cmd/release/templates/pick-sha.txt b/pkg/cmd/release/templates/pick-sha.txt | |
| deleted file mode 100644 | |
| index 2580241c8c..0000000000 | |
| --- a/pkg/cmd/release/templates/pick-sha.txt | |
| +++ /dev/null | |
| @@ -1,8 +0,0 @@ | |
| -A candidate SHA has been selected for {{ .Version }}. Proceeding to qualification shortly. | |
| - | |
| - SHA: {{ .SHA }} | |
| - Tracking Issue: {{ .TrackingIssueURL }} | |
| - List of changes since last release: {{ .DiffURL }} | |
| - | |
| -Thanks | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/templates/post-blockers.alpha.gohtml b/pkg/cmd/release/templates/post-blockers.alpha.gohtml | |
| deleted file mode 100644 | |
| index dd214c4268..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.alpha.gohtml | |
| +++ /dev/null | |
| @@ -1,54 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>{{ .Args.Version }}</strong> release is scheduled for prep & qualification on | |
| - <strong>{{ .Args.PrepDate }}</strong> for a scheduled release of <strong>{{ .Args.ReleaseDate }}</strong>. | |
| - </p> | |
| - <!-- ----- current blocker status and next steps ----- --> | |
| - {{ if (eq .Args.TotalBlockers 0) }} | |
| - <p> | |
| - There are currently <strong>0 open release blockers</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>) <span font-size='70%%'>🎉</span>. | |
| - We are clear to proceed with preparation and qualification. | |
| - </p> | |
| - {{ else }} | |
| - <p> | |
| - <strong>Respective teams: please review blockers below</strong> to assess if any of these cause | |
| - known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday | |
| - <strong>even if there are open release blockers</strong>. | |
| - </p> | |
| - <!-- grammar-specific conditional ("there is" vs "there are") --> | |
| - {{ if (eq .Args.TotalBlockers 1) }} | |
| - <p> | |
| - There is currently <strong>1 open release blocker</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>): | |
| - </p> | |
| - {{ else }} | |
| - <p> | |
| - There are currently <strong>{{ .Args.TotalBlockers }} open release blockers</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>): | |
| - </p> | |
| - {{ end }} | |
| - <ul> | |
| - {{ range .Args.BlockerList }} | |
| - <li>{{ .ProjectName }}: {{ .NumBlockers }}</li> | |
| - {{ end }} | |
| - </ul> | |
| - <p> | |
| - [1] As per our <a href='https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions' | |
| - >terms & definitions</a>, alphas/betas <strong>should not cause data loss or corruption</strong>, | |
| - but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - <ul> | |
| - <li>there would be no expectation that it would be used in prod or needs to be supported</li> | |
| - <li>alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release</li> | |
| - <li>alphas/betas may contain both known and unknown issues including speed or performance issues</li> | |
| - </ul> | |
| - </p> | |
| - {{ end }} | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/templates/post-blockers.alpha.subject b/pkg/cmd/release/templates/post-blockers.alpha.subject | |
| deleted file mode 100644 | |
| index e7681e6323..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.alpha.subject | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release {{ .Args.Version }} | |
| diff --git a/pkg/cmd/release/templates/post-blockers.alpha.txt b/pkg/cmd/release/templates/post-blockers.alpha.txt | |
| deleted file mode 100644 | |
| index 7ae158e591..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.alpha.txt | |
| +++ /dev/null | |
| @@ -1,36 +0,0 @@ | |
| -Hello! | |
| - | |
| -The {{ .Args.Version }} release is scheduled for prep & qualification on {{ .Args.PrepDate }} for a scheduled release of {{ .Args.ReleaseDate }}. | |
| -{{ | |
| - if (eq .Args.TotalBlockers 0) | |
| -}} | |
| -There are currently 0 open release blockers \o/ (https://{{ .Args.BlockersURL }}). We are clear to proceed with preparation and qualification. | |
| -{{ | |
| - else | |
| -}} | |
| -Respective teams: please review blockers below to assess if any of these cause known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday even if there are open release blockers. | |
| -{{ | |
| - if (eq .Args.TotalBlockers 1) | |
| -}} | |
| -There is currently 1 open release blocker (https://{{ .Args.BlockersURL }}):{{ | |
| - else | |
| -}} | |
| -There are currently {{ .Args.TotalBlockers }} open release blockers (https://{{ .Args.BlockersURL }}):{{ | |
| - end | |
| -}} | |
| -{{ | |
| - range .Args.BlockerList | |
| -}} - {{ .ProjectName }}: {{ .NumBlockers }} | |
| -{{ | |
| - end | |
| -}} | |
| -[1] As per our terms & definitions, alphas/betas should not cause data loss or corruption, but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - - there would be no expectation that it would be used in prod or needs to be supported | |
| - - alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release | |
| - - alphas/betas may contain both known and unknown issues including speed or performance issues | |
| -See our Release Terms & Conditions: https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions | |
| -{{ | |
| - end | |
| -}} | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/templates/post-blockers.gohtml b/pkg/cmd/release/templates/post-blockers.gohtml | |
| deleted file mode 100644 | |
| index a0ad43f58a..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.gohtml | |
| +++ /dev/null | |
| @@ -1,57 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>{{ .Args.Version }}</strong> release is scheduled for prep & qualification on | |
| - <strong>{{ .Args.PrepDate }}</strong> for a scheduled release of <strong>{{ .Args.ReleaseDate }}</strong>. | |
| - </p> | |
| - <!-- ----- backport direction, which depends on release-type ----- --> | |
| - {{ if .Template.BackportsUseBackboard }} | |
| - <p> | |
| - This is a reminder to merge any outstanding backports that need to be included in the next release. | |
| - Please refer to | |
| - <a href='https://backboard.crdb.dev/?branch={{ .Args.ReleaseBranch }}&repo=386372623925772289' | |
| - >backboard</a> to identify any such backports. | |
| - </p> | |
| - {{ end }} | |
| - {{ if .Template.BackportsWeeklyTriageReview }} | |
| - <p> | |
| - With the | |
| - <a href='https://go.crdb.dev/backport-policy-branch-cut'>Restrictive Backport Policy</a> in effect, | |
| - backports for {{ .Args.ReleaseBranch }} will be reviewed at Monday's triage meeting. | |
| - </p> | |
| - {{ end }} | |
| - <!-- ----- current blocker status and next steps ----- --> | |
| - {{ if (eq .Args.TotalBlockers 0) }} | |
| - <p> | |
| - There are currently <strong>0 open release blockers</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>) <span font-size='70%%'>🎉</span>. | |
| - We are clear to proceed with preparation and qualification. | |
| - </p> | |
| - {{ else }} | |
| - <!-- grammar-specific conditional ("there is" vs "there are") --> | |
| - {{ if (eq .Args.TotalBlockers 1) }} | |
| - <p> | |
| - There is currently <strong>1 open release blocker</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>), | |
| - which must be resolved before a candidate is chosen: | |
| - </p> | |
| - {{ else }} | |
| - <p> | |
| - There are currently <strong>{{ .Args.TotalBlockers }} open release blockers</strong> | |
| - (<a href='https://{{ .Args.BlockersURL }}'>{{ .Args.BlockersURL }}</a>), | |
| - which must be resolved before a candidate is chosen: | |
| - </p> | |
| - {{ end }} | |
| - <ul> | |
| - {{ range .Args.BlockerList }} | |
| - <li>{{ .ProjectName }}: {{ .NumBlockers }}</li> | |
| - {{ end }} | |
| - </ul> | |
| - {{ end }} | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/templates/post-blockers.subject b/pkg/cmd/release/templates/post-blockers.subject | |
| deleted file mode 100644 | |
| index e7681e6323..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.subject | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release {{ .Args.Version }} | |
| diff --git a/pkg/cmd/release/templates/post-blockers.txt b/pkg/cmd/release/templates/post-blockers.txt | |
| deleted file mode 100644 | |
| index a6c2a932ef..0000000000 | |
| --- a/pkg/cmd/release/templates/post-blockers.txt | |
| +++ /dev/null | |
| @@ -1,40 +0,0 @@ | |
| -Hello! | |
| - | |
| -The {{ .Args.Version }} release is scheduled for prep & qualification on {{ .Args.PrepDate }} for a scheduled release of {{ .Args.ReleaseDate }}. | |
| -{{ | |
| - if .Template.BackportsUseBackboard | |
| -}} | |
| -This is a reminder to merge any outstanding backports that need to be included in the next release. Please refer to backboard to identify any such backports, https://backboard.crdb.dev/?branch={{ .Args.ReleaseBranch }}&repo=386372623925772289 | |
| -{{ | |
| - end | |
| -}}{{ | |
| - if .Template.BackportsWeeklyTriageReview | |
| -}} | |
| -With the Restrictive Backport Policy in effect, backports for {{ .Args.ReleaseBranch }} will be reviewed at Monday's triage meeting. Backport policy: https://go.crdb.dev/backport-policy-branch-cut. | |
| -{{ | |
| - end | |
| -}}{{ | |
| - if (eq .Args.TotalBlockers 0) | |
| -}} | |
| -There are currently 0 open release blockers \o/ (https://{{ .Args.BlockersURL }}). We are clear to proceed with preparation and qualification. | |
| -{{ | |
| - else | |
| -}}{{ | |
| - if (eq .Args.TotalBlockers 1) | |
| -}} | |
| -There is currently 1 open release blocker (https://{{ .Args.BlockersURL }}), which must be resolved before a candidate is chosen:{{ | |
| - else | |
| -}} | |
| -There are currently {{ .Args.TotalBlockers }} open release blockers (https://{{ .Args.BlockersURL }}), which must be resolved before a candidate is chosen:{{ | |
| - end | |
| -}} | |
| -{{ | |
| - range .Args.BlockerList | |
| -}} - {{ .ProjectName }}: {{ .NumBlockers }} | |
| -{{ | |
| - end | |
| -}}{{ | |
| - end | |
| -}} | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/pick-sha.html.golden b/pkg/cmd/release/testdata/pick-sha.html.golden | |
| deleted file mode 100644 | |
| index 7dd9512497..0000000000 | |
| --- a/pkg/cmd/release/testdata/pick-sha.html.golden | |
| +++ /dev/null | |
| @@ -1,14 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>A candidate SHA has been selected for <strong>v21.1.13</strong>. Proceeding to qualification shortly.</p> | |
| - <ul> | |
| - <li>SHA: <a href="https://github.com/cockroachlabs/release-staging/commit/0fd6eead6c6eb7b2529deb39197cc3c95e93ded8">0fd6eead6c6eb7b2529deb39197cc3c95e93ded8</a></li> | |
| - <li>Tracking Issue: <a href="https://cockroachlabs.atlassian.net/browse/REL-111">REL-111</a></li> | |
| - <li><a href="https://github.com/cockroachdb/cockroach/compare/v21.1.13...0fd6eead6c6eb7b2529deb39197cc3c95e93ded8">List of changes</a> since last release</li> | |
| - </ul> | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/pick-sha.subject.golden b/pkg/cmd/release/testdata/pick-sha.subject.golden | |
| deleted file mode 100644 | |
| index f9ca091ed1..0000000000 | |
| --- a/pkg/cmd/release/testdata/pick-sha.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v21.1.13 | |
| diff --git a/pkg/cmd/release/testdata/pick-sha.txt.golden b/pkg/cmd/release/testdata/pick-sha.txt.golden | |
| deleted file mode 100644 | |
| index 9a3651a945..0000000000 | |
| --- a/pkg/cmd/release/testdata/pick-sha.txt.golden | |
| +++ /dev/null | |
| @@ -1,8 +0,0 @@ | |
| -A candidate SHA has been selected for v21.1.13. Proceeding to qualification shortly. | |
| - | |
| - SHA: 0fd6eead6c6eb7b2529deb39197cc3c95e93ded8 | |
| - Tracking Issue: https://cockroachlabs.atlassian.net/browse/REL-111 | |
| - List of changes since last release: https://github.com/cockroachdb/cockroach/compare/v21.1.13...0fd6eead6c6eb7b2529deb39197cc3c95e93ded8 | |
| - | |
| -Thanks | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.1-blocker.html.golden b/pkg/cmd/release/testdata/post-blockers.1-blocker.html.golden | |
| deleted file mode 100644 | |
| index b746358fd0..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.1-blocker.html.golden | |
| +++ /dev/null | |
| @@ -1,38 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.0-beta.2</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - | |
| - <p> | |
| - With the | |
| - <a href='https://go.crdb.dev/backport-policy-branch-cut'>Restrictive Backport Policy</a> in effect, | |
| - backports for master will be reviewed at Monday's triage meeting. | |
| - </p> | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - <p> | |
| - There is currently <strong>1 open release blocker</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>), | |
| - which must be resolved before a candidate is chosen: | |
| - </p> | |
| - | |
| - <ul> | |
| - | |
| - <li>Project 1: 1</li> | |
| - | |
| - </ul> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.1-blocker.subject.golden b/pkg/cmd/release/testdata/post-blockers.1-blocker.subject.golden | |
| deleted file mode 100644 | |
| index e168458a12..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.1-blocker.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.0-beta.2 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.1-blocker.txt.golden b/pkg/cmd/release/testdata/post-blockers.1-blocker.txt.golden | |
| deleted file mode 100644 | |
| index 6049f4dc92..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.1-blocker.txt.golden | |
| +++ /dev/null | |
| @@ -1,11 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.0-beta.2 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -With the Restrictive Backport Policy in effect, backports for master will be reviewed at Monday's triage meeting. Backport policy: https://go.crdb.dev/backport-policy-branch-cut. | |
| - | |
| -There is currently 1 open release blocker (https://go.crdb.dev/blockers), which must be resolved before a candidate is chosen: | |
| - - Project 1: 1 | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.html.golden b/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.html.golden | |
| deleted file mode 100644 | |
| index c5f6a06944..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.html.golden | |
| +++ /dev/null | |
| @@ -1,43 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.0-alpha.3</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - <strong>Respective teams: please review blockers below</strong> to assess if any of these cause | |
| - known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday | |
| - <strong>even if there are open release blockers</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - There is currently <strong>1 open release blocker</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>): | |
| - </p> | |
| - | |
| - <ul> | |
| - | |
| - <li>Project 1: 1</li> | |
| - | |
| - </ul> | |
| - <p> | |
| - [1] As per our <a href='https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions' | |
| - >terms & definitions</a>, alphas/betas <strong>should not cause data loss or corruption</strong>, | |
| - but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - <ul> | |
| - <li>there would be no expectation that it would be used in prod or needs to be supported</li> | |
| - <li>alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release</li> | |
| - <li>alphas/betas may contain both known and unknown issues including speed or performance issues</li> | |
| - </ul> | |
| - </p> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.subject.golden b/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.subject.golden | |
| deleted file mode 100644 | |
| index ff4d866600..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.0-alpha.3 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.txt.golden b/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.txt.golden | |
| deleted file mode 100644 | |
| index b61bde0964..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.1-blocker.txt.golden | |
| +++ /dev/null | |
| @@ -1,17 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.0-alpha.3 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -Respective teams: please review blockers below to assess if any of these cause known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday even if there are open release blockers. | |
| - | |
| -There is currently 1 open release blocker (https://go.crdb.dev/blockers): | |
| - - Project 1: 1 | |
| - | |
| -[1] As per our terms & definitions, alphas/betas should not cause data loss or corruption, but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - - there would be no expectation that it would be used in prod or needs to be supported | |
| - - alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release | |
| - - alphas/betas may contain both known and unknown issues including speed or performance issues | |
| -See our Release Terms & Conditions: https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.html.golden b/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.html.golden | |
| deleted file mode 100644 | |
| index 9b29adb3be..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.html.golden | |
| +++ /dev/null | |
| @@ -1,47 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.0-alpha.3</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - <strong>Respective teams: please review blockers below</strong> to assess if any of these cause | |
| - known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday | |
| - <strong>even if there are open release blockers</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - There are currently <strong>9 open release blockers</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>): | |
| - </p> | |
| - | |
| - <ul> | |
| - | |
| - <li>Project 1: 4</li> | |
| - | |
| - <li>Project 2: 2</li> | |
| - | |
| - <li>Project 3: 3</li> | |
| - | |
| - </ul> | |
| - <p> | |
| - [1] As per our <a href='https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions' | |
| - >terms & definitions</a>, alphas/betas <strong>should not cause data loss or corruption</strong>, | |
| - but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - <ul> | |
| - <li>there would be no expectation that it would be used in prod or needs to be supported</li> | |
| - <li>alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release</li> | |
| - <li>alphas/betas may contain both known and unknown issues including speed or performance issues</li> | |
| - </ul> | |
| - </p> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.subject.golden b/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.subject.golden | |
| deleted file mode 100644 | |
| index ff4d866600..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.0-alpha.3 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.txt.golden b/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.txt.golden | |
| deleted file mode 100644 | |
| index cf990c7331..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.many-blockers.txt.golden | |
| +++ /dev/null | |
| @@ -1,19 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.0-alpha.3 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -Respective teams: please review blockers below to assess if any of these cause known data loss or corruption[1]. Otherwise, we will proceed to prep & qualify on Tuesday even if there are open release blockers. | |
| - | |
| -There are currently 9 open release blockers (https://go.crdb.dev/blockers): | |
| - - Project 1: 4 | |
| - - Project 2: 2 | |
| - - Project 3: 3 | |
| - | |
| -[1] As per our terms & definitions, alphas/betas should not cause data loss or corruption, but may contain both known and unknown issues including speed or performance issues. As with other alphas: | |
| - - there would be no expectation that it would be used in prod or needs to be supported | |
| - - alpha clusters must be wiped, and cannot be upgraded to 22.1 betas, RCs or final production release | |
| - - alphas/betas may contain both known and unknown issues including speed or performance issues | |
| -See our Release Terms & Conditions: https://cockroachlabs.atlassian.net/wiki/spaces/ENG/pages/869990521/Release+Process+Terms+and+Definitions | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.html.golden b/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.html.golden | |
| deleted file mode 100644 | |
| index 5c6115f1de..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.html.golden | |
| +++ /dev/null | |
| @@ -1,21 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.0-alpha.3</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - There are currently <strong>0 open release blockers</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>) <span font-size='70%%'>🎉</span>. | |
| - We are clear to proceed with preparation and qualification. | |
| - </p> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.subject.golden b/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.subject.golden | |
| deleted file mode 100644 | |
| index ff4d866600..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.0-alpha.3 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.txt.golden b/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.txt.golden | |
| deleted file mode 100644 | |
| index 428fcad1d5..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.alpha.zero-blockers.txt.golden | |
| +++ /dev/null | |
| @@ -1,8 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.0-alpha.3 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -There are currently 0 open release blockers \o/ (https://go.crdb.dev/blockers). We are clear to proceed with preparation and qualification. | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.many-blockers.html.golden b/pkg/cmd/release/testdata/post-blockers.many-blockers.html.golden | |
| deleted file mode 100644 | |
| index 1be3d57191..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.many-blockers.html.golden | |
| +++ /dev/null | |
| @@ -1,42 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.0-rc.3</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - | |
| - <p> | |
| - With the | |
| - <a href='https://go.crdb.dev/backport-policy-branch-cut'>Restrictive Backport Policy</a> in effect, | |
| - backports for master will be reviewed at Monday's triage meeting. | |
| - </p> | |
| - | |
| - | |
| - | |
| - | |
| - | |
| - <p> | |
| - There are currently <strong>9 open release blockers</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>), | |
| - which must be resolved before a candidate is chosen: | |
| - </p> | |
| - | |
| - <ul> | |
| - | |
| - <li>Project 1: 4</li> | |
| - | |
| - <li>Project 2: 2</li> | |
| - | |
| - <li>Project 3: 3</li> | |
| - | |
| - </ul> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.many-blockers.subject.golden b/pkg/cmd/release/testdata/post-blockers.many-blockers.subject.golden | |
| deleted file mode 100644 | |
| index 3c0e921ce0..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.many-blockers.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.0-rc.3 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.many-blockers.txt.golden b/pkg/cmd/release/testdata/post-blockers.many-blockers.txt.golden | |
| deleted file mode 100644 | |
| index e8ed81da2f..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.many-blockers.txt.golden | |
| +++ /dev/null | |
| @@ -1,13 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.0-rc.3 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -With the Restrictive Backport Policy in effect, backports for master will be reviewed at Monday's triage meeting. Backport policy: https://go.crdb.dev/backport-policy-branch-cut. | |
| - | |
| -There are currently 9 open release blockers (https://go.crdb.dev/blockers), which must be resolved before a candidate is chosen: | |
| - - Project 1: 4 | |
| - - Project 2: 2 | |
| - - Project 3: 3 | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.zero-blockers.html.golden b/pkg/cmd/release/testdata/post-blockers.zero-blockers.html.golden | |
| deleted file mode 100644 | |
| index 75e04ac841..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.zero-blockers.html.golden | |
| +++ /dev/null | |
| @@ -1,31 +0,0 @@ | |
| -<html> | |
| - <body> | |
| - <p>Hello!</p> | |
| - <p> | |
| - The <strong>v19.1.11</strong> release is scheduled for prep & qualification on | |
| - <strong>Saturday, April 1</strong> for a scheduled release of <strong>Saturday, April 11</strong>. | |
| - </p> | |
| - | |
| - | |
| - <p> | |
| - This is a reminder to merge any outstanding backports that need to be included in the next release. | |
| - Please refer to | |
| - <a href='https://backboard.crdb.dev/?branch=master&repo=386372623925772289' | |
| - >backboard</a> to identify any such backports. | |
| - </p> | |
| - | |
| - | |
| - | |
| - | |
| - <p> | |
| - There are currently <strong>0 open release blockers</strong> | |
| - (<a href='https://go.crdb.dev/blockers'>go.crdb.dev/blockers</a>) <span font-size='70%%'>🎉</span>. | |
| - We are clear to proceed with preparation and qualification. | |
| - </p> | |
| - | |
| - <p> | |
| - Thanks,<br /> | |
| - Release Engineering | |
| - </p> | |
| - </body> | |
| -</html> | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.zero-blockers.subject.golden b/pkg/cmd/release/testdata/post-blockers.zero-blockers.subject.golden | |
| deleted file mode 100644 | |
| index bd85bcf500..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.zero-blockers.subject.golden | |
| +++ /dev/null | |
| @@ -1 +0,0 @@ | |
| -Release v19.1.11 | |
| diff --git a/pkg/cmd/release/testdata/post-blockers.zero-blockers.txt.golden b/pkg/cmd/release/testdata/post-blockers.zero-blockers.txt.golden | |
| deleted file mode 100644 | |
| index 87f7181424..0000000000 | |
| --- a/pkg/cmd/release/testdata/post-blockers.zero-blockers.txt.golden | |
| +++ /dev/null | |
| @@ -1,10 +0,0 @@ | |
| -Hello! | |
| - | |
| -The v19.1.11 release is scheduled for prep & qualification on Saturday, April 1 for a scheduled release of Saturday, April 11. | |
| - | |
| -This is a reminder to merge any outstanding backports that need to be included in the next release. Please refer to backboard to identify any such backports, https://backboard.crdb.dev/?branch=master&repo=386372623925772289 | |
| - | |
| -There are currently 0 open release blockers \o/ (https://go.crdb.dev/blockers). We are clear to proceed with preparation and qualification. | |
| - | |
| -Thanks, | |
| -Release Engineering | |
| diff --git a/pkg/cmd/roachtest/BUILD.bazel b/pkg/cmd/roachtest/BUILD.bazel | |
| index 9f6fdf8fd1..79eef8d40c 100644 | |
| --- a/pkg/cmd/roachtest/BUILD.bazel | |
| +++ b/pkg/cmd/roachtest/BUILD.bazel | |
| @@ -74,7 +74,6 @@ go_test( | |
| "//pkg/cmd/roachtest/test", | |
| "//pkg/roachprod/logger", | |
| "//pkg/testutils", | |
| - "//pkg/util/quotapool", | |
| "//pkg/util/stop", | |
| "//pkg/util/syncutil", | |
| "//pkg/util/version", | |
| diff --git a/pkg/cmd/roachtest/cluster.go b/pkg/cmd/roachtest/cluster.go | |
| index 5abf38d35d..c4e867b37c 100644 | |
| --- a/pkg/cmd/roachtest/cluster.go | |
| +++ b/pkg/cmd/roachtest/cluster.go | |
| @@ -818,15 +818,6 @@ func createFlagsOverride(flags *pflag.FlagSet, opts *vm.CreateOpts) { | |
| } | |
| } | |
| -// clusterMock creates a cluster to be used for (self) testing. | |
| -func (f *clusterFactory) clusterMock(cfg clusterConfig) *clusterImpl { | |
| - return &clusterImpl{ | |
| - name: f.genName(cfg), | |
| - expiration: timeutil.Now().Add(24 * time.Hour), | |
| - r: f.r, | |
| - } | |
| -} | |
| - | |
| // newCluster creates a new roachprod cluster. | |
| // | |
| // setStatus is called with status messages indicating the stage of cluster | |
| @@ -845,8 +836,12 @@ func (f *clusterFactory) newCluster( | |
| } | |
| if cfg.spec.NodeCount == 0 { | |
| - // For tests, use a mock cluster. | |
| - c := f.clusterMock(cfg) | |
| + // For tests. Return the minimum that makes them happy. | |
| + c := &clusterImpl{ | |
| + name: f.genName(cfg), | |
| + expiration: timeutil.Now().Add(24 * time.Hour), | |
| + r: f.r, | |
| + } | |
| if err := f.r.registerCluster(c); err != nil { | |
| return nil, err | |
| } | |
| @@ -1367,7 +1362,7 @@ WHERE t.status NOT IN ('RANGE_CONSISTENT', 'RANGE_INDETERMINATE')`) | |
| // crdb_internal.check_consistency(true, '', '') indicates that any ranges' | |
| // replicas are inconsistent with each other. It uses the first node that | |
| // is up to run the query. | |
| -func (c *clusterImpl) FailOnReplicaDivergence(ctx context.Context, t *testImpl) { | |
| +func (c *clusterImpl) FailOnReplicaDivergence(ctx context.Context, t test.Test) { | |
| if c.spec.NodeCount < 1 { | |
| return // unit tests | |
| } | |
| diff --git a/pkg/cmd/roachtest/main.go b/pkg/cmd/roachtest/main.go | |
| index fdbe9c817a..28eca72c22 100644 | |
| --- a/pkg/cmd/roachtest/main.go | |
| +++ b/pkg/cmd/roachtest/main.go | |
| @@ -415,7 +415,7 @@ func runTests(register func(registry.Registry), cfg cliCfg) error { | |
| err = runner.Run( | |
| ctx, tests, cfg.count, cfg.parallelism, opt, | |
| testOpts{versionsBinaryOverride: cfg.versionsBinaryOverride}, | |
| - lopt, nil /* clusterAllocator */) | |
| + lopt) | |
| // Make sure we attempt to clean up. We run with a non-canceled ctx; the | |
| // ctx above might be canceled in case a signal was received. If that's | |
| diff --git a/pkg/cmd/roachtest/prometheus/BUILD.bazel b/pkg/cmd/roachtest/prometheus/BUILD.bazel | |
| index bcdf3635b8..3046155493 100644 | |
| --- a/pkg/cmd/roachtest/prometheus/BUILD.bazel | |
| +++ b/pkg/cmd/roachtest/prometheus/BUILD.bazel | |
| @@ -9,8 +9,6 @@ go_library( | |
| deps = [ | |
| "//pkg/cmd/roachtest/option", | |
| "//pkg/roachprod/logger", | |
| - "@com_github_prometheus_client_golang//api/prometheus/v1:prometheus", | |
| - "@com_github_prometheus_common//model", | |
| "@in_gopkg_yaml_v2//:yaml_v2", | |
| ], | |
| ) | |
| diff --git a/pkg/cmd/roachtest/prometheus/prometheus.go b/pkg/cmd/roachtest/prometheus/prometheus.go | |
| index dda5b932ca..7f0d520713 100644 | |
| --- a/pkg/cmd/roachtest/prometheus/prometheus.go | |
| +++ b/pkg/cmd/roachtest/prometheus/prometheus.go | |
| @@ -21,16 +21,9 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/cmd/roachtest/option" | |
| "github.com/cockroachdb/cockroach/pkg/roachprod/logger" | |
| - promv1 "github.com/prometheus/client_golang/api/prometheus/v1" | |
| - "github.com/prometheus/common/model" | |
| "gopkg.in/yaml.v2" | |
| ) | |
| -// Client is an interface allowing queries against Prometheus. | |
| -type Client interface { | |
| - Query(ctx context.Context, query string, ts time.Time) (model.Value, promv1.Warnings, error) | |
| -} | |
| - | |
| // ScrapeNode are nodes to scrape from. | |
| type ScrapeNode struct { | |
| Nodes option.NodeListOption | |
| diff --git a/pkg/cmd/roachtest/registry/owners.go b/pkg/cmd/roachtest/registry/owners.go | |
| index 8f3e61822f..c0f2c6bce9 100644 | |
| --- a/pkg/cmd/roachtest/registry/owners.go | |
| +++ b/pkg/cmd/roachtest/registry/owners.go | |
| @@ -27,5 +27,4 @@ const ( | |
| OwnerSQLSchema Owner = `sql-schema` | |
| OwnerStorage Owner = `storage` | |
| OwnerTestEng Owner = `test-eng` | |
| - OwnerDevInf Owner = `dev-inf` | |
| ) | |
| diff --git a/pkg/cmd/roachtest/test_runner.go b/pkg/cmd/roachtest/test_runner.go | |
| index 877022769e..9655eba8f7 100644 | |
| --- a/pkg/cmd/roachtest/test_runner.go | |
| +++ b/pkg/cmd/roachtest/test_runner.go | |
| @@ -20,13 +20,13 @@ import ( | |
| "net" | |
| "net/http" | |
| "os" | |
| + "os/exec" | |
| "path/filepath" | |
| "runtime" | |
| "sort" | |
| "strconv" | |
| "strings" | |
| "sync" | |
| - "sync/atomic" | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/cmd/internal/issues" | |
| @@ -50,7 +50,6 @@ import ( | |
| ) | |
| var errTestsFailed = fmt.Errorf("some tests failed") | |
| -var errClusterProvisioningFailed = fmt.Errorf("some clusters could not be created") | |
| // testRunner runs tests. | |
| type testRunner struct { | |
| @@ -95,9 +94,6 @@ type testRunner struct { | |
| // completed maintains information on all completed test runs. | |
| completed []completedTestInfo | |
| } | |
| - | |
| - // Counts cluster creation errors across all workers. | |
| - numClusterErrs int32 | |
| } | |
| // newTestRunner constructs a testRunner. | |
| @@ -181,7 +177,6 @@ func (r *testRunner) Run( | |
| clustersOpt clustersOpt, | |
| topt testOpts, | |
| lopt loggingOpt, | |
| - clusterAllocator clusterAllocatorFn, | |
| ) error { | |
| // Validate options. | |
| if len(tests) == 0 { | |
| @@ -225,8 +220,58 @@ func (r *testRunner) Run( | |
| } | |
| } | |
| } | |
| - if clusterAllocator == nil { | |
| - clusterAllocator = defaultClusterAllocator(r, clustersOpt, lopt) | |
| + | |
| + var numConcurrentClusterCreations int | |
| + if cloud == "aws" { | |
| + // AWS has ridiculous API calls limits, so we're going to create one cluster | |
| + // at a time. Internally, roachprod has throttling for the calls required to | |
| + // create a single cluster. | |
| + numConcurrentClusterCreations = 1 | |
| + } else { | |
| + numConcurrentClusterCreations = 1000 | |
| + } | |
| + clusterFactory := newClusterFactory( | |
| + clustersOpt.user, clustersOpt.clusterID, lopt.artifactsDir, r.cr, numConcurrentClusterCreations) | |
| + | |
| + // allocateCluster will be used by workers to create new clusters (or to attach | |
| + // to an existing one). | |
| + allocateCluster := func( | |
| + ctx context.Context, | |
| + t registry.TestSpec, | |
| + alloc *quotapool.IntAlloc, | |
| + artifactsDir string, | |
| + wStatus *workerStatus, | |
| + ) (*clusterImpl, error) { | |
| + wStatus.SetStatus("creating cluster") | |
| + defer wStatus.SetStatus("") | |
| + | |
| + lopt.l.PrintfCtx(ctx, "Creating new cluster for test %s: %s", t.Name, t.Cluster) | |
| + | |
| + existingClusterName := clustersOpt.clusterName | |
| + if existingClusterName != "" { | |
| + // Logs for attaching to a cluster go to a dedicated log file. | |
| + logPath := filepath.Join(artifactsDir, runnerLogsDir, "cluster-create", existingClusterName+".log") | |
| + clusterL, err := logger.RootLogger(logPath, lopt.tee) | |
| + if err != nil { | |
| + return nil, err | |
| + } | |
| + defer clusterL.Close() | |
| + opt := attachOpt{ | |
| + skipValidation: r.config.skipClusterValidationOnAttach, | |
| + skipStop: r.config.skipClusterStopOnAttach, | |
| + skipWipe: r.config.skipClusterWipeOnAttach, | |
| + } | |
| + return attachToExistingCluster(ctx, existingClusterName, clusterL, t.Cluster, opt, r.cr) | |
| + } | |
| + | |
| + cfg := clusterConfig{ | |
| + spec: t.Cluster, | |
| + artifactsDir: artifactsDir, | |
| + username: clustersOpt.user, | |
| + localCluster: clustersOpt.typ == localCluster, | |
| + alloc: alloc, | |
| + } | |
| + return clusterFactory.newCluster(ctx, cfg, wStatus.SetStatus, lopt.tee) | |
| } | |
| // Seed the default rand source so that different runs get different cluster | |
| @@ -249,25 +294,23 @@ func (r *testRunner) Run( | |
| qp := quotapool.NewIntPool("cloud cpu", uint64(clustersOpt.cpuQuota)) | |
| l := lopt.l | |
| - var wg sync.WaitGroup | |
| + var wg sync.WaitGroup | |
| for i := 0; i < parallelism; i++ { | |
| i := i // Copy for closure. | |
| wg.Add(1) | |
| if err := r.stopper.RunAsyncTask(ctx, "worker", func(ctx context.Context) { | |
| defer wg.Done() | |
| - err := r.runWorker( | |
| + if err := r.runWorker( | |
| ctx, fmt.Sprintf("w%d", i) /* name */, r.work, qp, | |
| r.stopper.ShouldQuiesce(), | |
| clustersOpt.keepClustersOnTestFailure, | |
| - lopt.artifactsDir, lopt.literalArtifactsDir, lopt.tee, lopt.stdout, | |
| - clusterAllocator, | |
| + lopt.artifactsDir, lopt.literalArtifactsDir, lopt.runnerLogPath, lopt.tee, lopt.stdout, | |
| + allocateCluster, | |
| topt, | |
| l, | |
| - ) | |
| - | |
| - if err != nil { | |
| + ); err != nil { | |
| // A worker returned an error. Let's shut down. | |
| msg := fmt.Sprintf("Worker %d returned with error. Quiescing. Error: %v", i, err) | |
| shout(ctx, l, lopt.stdout, msg) | |
| @@ -300,83 +343,12 @@ func (r *testRunner) Run( | |
| } | |
| passFailLine := r.generateReport() | |
| shout(ctx, l, lopt.stdout, passFailLine) | |
| - | |
| - if r.numClusterErrs > 0 { | |
| - shout(ctx, l, lopt.stdout, "%d clusters could not be created", r.numClusterErrs) | |
| - return errClusterProvisioningFailed | |
| - } | |
| - | |
| if len(r.status.fail) > 0 { | |
| return errTestsFailed | |
| } | |
| return nil | |
| } | |
| -// N.B. currently this value is hardcoded per cloud provider. | |
| -func numConcurrentClusterCreations() int { | |
| - var res int | |
| - if cloud == "aws" { | |
| - // AWS has ridiculous API calls limits, so we're going to create one cluster | |
| - // at a time. Internally, roachprod has throttling for the calls required to | |
| - // create a single cluster. | |
| - res = 1 | |
| - } else { | |
| - res = 1000 | |
| - } | |
| - return res | |
| -} | |
| - | |
| -// defaultClusterAllocator is used by workers to create new clusters (or to attach | |
| -// to an existing one). | |
| -// | |
| -// N.B. the resulting clusterAllocatorFn reuses the same clusterFactory to allocate clusters. | |
| -func defaultClusterAllocator( | |
| - r *testRunner, clustersOpt clustersOpt, lopt loggingOpt, | |
| -) clusterAllocatorFn { | |
| - clusterFactory := newClusterFactory( | |
| - clustersOpt.user, clustersOpt.clusterID, lopt.artifactsDir, r.cr, numConcurrentClusterCreations()) | |
| - | |
| - allocateCluster := func( | |
| - ctx context.Context, | |
| - t registry.TestSpec, | |
| - alloc *quotapool.IntAlloc, | |
| - artifactsDir string, | |
| - wStatus *workerStatus, | |
| - ) (*clusterImpl, error) { | |
| - wStatus.SetStatus("creating cluster") | |
| - defer wStatus.SetStatus("") | |
| - | |
| - existingClusterName := clustersOpt.clusterName | |
| - if existingClusterName != "" { | |
| - // Logs for attaching to a cluster go to a dedicated log file. | |
| - logPath := filepath.Join(artifactsDir, runnerLogsDir, "cluster-create", existingClusterName+".log") | |
| - clusterL, err := logger.RootLogger(logPath, lopt.tee) | |
| - if err != nil { | |
| - return nil, err | |
| - } | |
| - defer clusterL.Close() | |
| - opt := attachOpt{ | |
| - skipValidation: r.config.skipClusterValidationOnAttach, | |
| - skipStop: r.config.skipClusterStopOnAttach, | |
| - skipWipe: r.config.skipClusterWipeOnAttach, | |
| - } | |
| - lopt.l.PrintfCtx(ctx, "Attaching to existing cluster %s for test %s", existingClusterName, t.Name) | |
| - return attachToExistingCluster(ctx, existingClusterName, clusterL, t.Cluster, opt, r.cr) | |
| - } | |
| - lopt.l.PrintfCtx(ctx, "Creating new cluster for test %s: %s", t.Name, t.Cluster) | |
| - | |
| - cfg := clusterConfig{ | |
| - spec: t.Cluster, | |
| - artifactsDir: artifactsDir, | |
| - username: clustersOpt.user, | |
| - localCluster: clustersOpt.typ == localCluster, | |
| - alloc: alloc, | |
| - } | |
| - return clusterFactory.newCluster(ctx, cfg, wStatus.SetStatus, lopt.tee) | |
| - } | |
| - return allocateCluster | |
| -} | |
| - | |
| type clusterAllocatorFn func( | |
| ctx context.Context, | |
| t registry.TestSpec, | |
| @@ -396,18 +368,14 @@ type clusterAllocatorFn func( | |
| // nonresponsive to timeout case) which might still be running and doing | |
| // arbitrary things to the cluster it was using. | |
| // | |
| -// If a cluster cannot be provisioned (owing to an infrastructure issue), the corresponding | |
| -// test is skipped; the provisioning error is posted to github; the count of cluster provisioning | |
| -// errors is incremented. | |
| -// | |
| -// runWorker returns either error (other than cluster provisioning) or the count of cluster provisioning errors. | |
| -// | |
| // Args: | |
| // name: The worker's name, to be used as a prefix for log messages. | |
| // artifactsRootDir: The artifacts dir. Each test's logs are going to be under a | |
| // run_<n> dir. If empty, test log files will not be created. | |
| // literalArtifactsDir: The literal on-agent path where artifacts are stored. | |
| // Only used for teamcity[publishArtifacts] messages. | |
| +// testRunnerLogPath: The path to the test runner's log. It will be copied to | |
| +// failing tests' artifacts dir if running under TeamCity. | |
| // stdout: The Writer to use for messages that need to go to stdout (e.g. the | |
| // "=== RUN" and "--- FAIL" lines). | |
| // teeOpt: The teeing option for future test loggers. | |
| @@ -421,6 +389,7 @@ func (r *testRunner) runWorker( | |
| debug bool, | |
| artifactsRootDir string, | |
| literalArtifactsDir string, | |
| + testRunnerLogPath string, | |
| teeOpt logger.TeeOptType, | |
| stdout io.Writer, | |
| allocateCluster clusterAllocatorFn, | |
| @@ -434,6 +403,7 @@ func (r *testRunner) runWorker( | |
| }() | |
| var c *clusterImpl // The cluster currently being used. | |
| + var err error | |
| // When this method returns we'll destroy the cluster we had at the time. | |
| // Note that, if debug was set, c has been set to nil. | |
| defer func() { | |
| @@ -453,6 +423,7 @@ func (r *testRunner) runWorker( | |
| l.PrintfCtx(ctx, "Worker exiting with canceled ctx. Not destroying cluster.") | |
| } | |
| }() | |
| + | |
| // Loop until there's no more work in the pool, we get interrupted, or an | |
| // error occurs. | |
| for { | |
| @@ -476,71 +447,24 @@ func (r *testRunner) runWorker( | |
| } | |
| } | |
| var testToRun testToRunRes | |
| - var err error | |
| - | |
| wStatus.SetTest(nil /* test */, testToRunRes{}) | |
| wStatus.SetStatus("getting work") | |
| - testToRun, err = r.getWork( | |
| + testToRun, c, err = r.getWork( | |
| ctx, work, qp, c, interrupt, l, | |
| getWorkCallbacks{ | |
| + createCluster: func(ctx context.Context, ttr testToRunRes) (*clusterImpl, error) { | |
| + wStatus.SetTest(nil /* test */, ttr) | |
| + return allocateCluster(ctx, ttr.spec, ttr.alloc, artifactsRootDir, wStatus) | |
| + }, | |
| onDestroy: func() { | |
| wStatus.SetCluster(nil) | |
| }, | |
| }) | |
| - if err != nil { | |
| - // Problem selecting a test, bail out. | |
| + if err != nil || testToRun.noWork { | |
| return err | |
| } | |
| - if testToRun.noWork { | |
| - shout(ctx, l, stdout, "no work remaining; runWorker is bailing out...") | |
| - return nil | |
| - } | |
| - // Attempt to reuse existing cluster. | |
| - if c != nil && testToRun.canReuseCluster { | |
| - err = func() error { | |
| - l.PrintfCtx(ctx, "Using existing cluster: %s. Wiping", c.name) | |
| - if err := c.WipeE(ctx, l); err != nil { | |
| - return err | |
| - } | |
| - if err := c.RunE(ctx, c.All(), "rm -rf "+perfArtifactsDir); err != nil { | |
| - return errors.Wrapf(err, "failed to remove perf artifacts dir") | |
| - } | |
| - if c.localCertsDir != "" { | |
| - if err := os.RemoveAll(c.localCertsDir); err != nil { | |
| - return errors.Wrapf(err, | |
| - "failed to remove local certs in %s", c.localCertsDir) | |
| - } | |
| - c.localCertsDir = "" | |
| - } | |
| - // Overwrite the spec of the cluster with the one coming from the test. In | |
| - // particular, this overwrites the reuse policy to reflect what the test | |
| - // intends to do with it. | |
| - c.spec = testToRun.spec.Cluster | |
| - return nil | |
| - }() | |
| - if err != nil { | |
| - // N.B. handle any error during reuse attempt as clusterCreateErr. | |
| - shout(ctx, l, stdout, "Unable to reuse cluster: %s due to: %s. Will attempt to create a fresh one", | |
| - c.Name(), err) | |
| - atomic.AddInt32(&r.numClusterErrs, 1) | |
| - // Let's attempt to create a fresh one. | |
| - testToRun.canReuseCluster = false | |
| - } | |
| - } | |
| - var clusterCreateErr error | |
| - | |
| - if !testToRun.canReuseCluster { | |
| - // Create a new cluster if can't reuse or reuse attempt failed. | |
| - // N.B. non-reusable cluster would have been destroyed above. | |
| - wStatus.SetTest(nil /* test */, testToRun) | |
| - wStatus.SetStatus("creating cluster") | |
| - c, clusterCreateErr = allocateCluster(ctx, testToRun.spec, testToRun.alloc, artifactsRootDir, wStatus) | |
| - if clusterCreateErr != nil { | |
| - atomic.AddInt32(&r.numClusterErrs, 1) | |
| - shout(ctx, l, stdout, "Unable to create (or reuse) cluster for test %s due to: %s.", | |
| - testToRun.spec.Name, clusterCreateErr) | |
| - } | |
| - } | |
| + c.status("running test") | |
| + | |
| // Prepare the test's logger. | |
| logPath := "" | |
| var artifactsDir string | |
| @@ -571,51 +495,28 @@ func (r *testRunner) runWorker( | |
| l: testL, | |
| versionsBinaryOverride: topt.versionsBinaryOverride, | |
| } | |
| + // Tell the cluster that, from now on, it will be run "on behalf of this | |
| + // test". | |
| + c.setTest(t) | |
| + wStatus.SetCluster(c) | |
| + wStatus.SetTest(t, testToRun) | |
| + wStatus.SetStatus("running test") | |
| + | |
| // Now run the test. | |
| l.PrintfCtx(ctx, "starting test: %s:%d", testToRun.spec.Name, testToRun.runNum) | |
| - | |
| - if clusterCreateErr != nil { | |
| - // N.B. cluster creation must have failed... | |
| - // We don't want to prematurely abort the test suite since it's likely a transient issue. | |
| - // Instead, let's report an infrastructure issue, mark the test as failed and continue with the next test. | |
| - // Note, we fake the test name so that all cluster creation errors are posted to the same github issue. | |
| - oldName := t.spec.Name | |
| - oldOwner := t.spec.Owner | |
| - // Generate failure reason and mark the test failed to preclude fetching (cluster) artifacts. | |
| - t.printAndFail(0, clusterCreateErr) | |
| - issueOutput := "test %s was skipped due to %s" | |
| - issueOutput = fmt.Sprintf(issueOutput, oldName, t.FailureMsg()) | |
| - // N.B. issue title is of the form "roachtest: ${t.spec.Name} failed" (see UnitTestFormatter). | |
| - t.spec.Name = "cluster_creation" | |
| - t.spec.Owner = registry.OwnerDevInf | |
| - r.maybePostGithubIssue(ctx, l, t, stdout, issueOutput) | |
| - // Restore test name and owner. | |
| - t.spec.Name = oldName | |
| - t.spec.Owner = oldOwner | |
| - } else { | |
| - // Tell the cluster that, from now on, it will be run "on behalf of this | |
| - // test". | |
| - c.status("running test") | |
| - c.setTest(t) | |
| - wStatus.SetCluster(c) | |
| - wStatus.SetTest(t, testToRun) | |
| - wStatus.SetStatus("running test") | |
| - | |
| - err = r.runTest(ctx, t, testToRun.runNum, testToRun.runCount, c, stdout, testL) | |
| - } | |
| - | |
| - if err != nil { | |
| + if err := r.runTest( | |
| + ctx, t, testToRun.runNum, testToRun.runCount, c, testRunnerLogPath, stdout, testL, | |
| + ); err != nil { | |
| shout(ctx, l, stdout, "test returned error: %s: %s", t.Name(), err) | |
| // Mark the test as failed if it isn't already. | |
| if !t.Failed() { | |
| t.printAndFail(0 /* skip */, err) | |
| } | |
| } else { | |
| - msg := "test passed: %s (run %d)" | |
| + msg := "test passed" | |
| if t.Failed() { | |
| - msg = "test failed: %s (run %d)" | |
| + msg = fmt.Sprintf("test failed: %s (run %d)", t.Name(), testToRun.runNum) | |
| } | |
| - msg = fmt.Sprintf(msg, t.Name(), testToRun.runNum) | |
| l.PrintfCtx(ctx, msg) | |
| } | |
| testL.Close() | |
| @@ -626,23 +527,22 @@ func (r *testRunner) runWorker( | |
| } else { | |
| failureMsg += t.FailureMsg() | |
| } | |
| - if c != nil { | |
| - if debug { | |
| - // Save the cluster for future debugging. | |
| - c.Save(ctx, failureMsg, l) | |
| - | |
| - // Continue with a fresh cluster. | |
| - c = nil | |
| - } else { | |
| - // On any test failure or error, we destroy the cluster. We could be | |
| - // more selective, but this sounds safer. | |
| - l.PrintfCtx(ctx, "destroying cluster %s because: %s", c, failureMsg) | |
| - c.Destroy(context.Background(), closeLogger, l) | |
| - c = nil | |
| - } | |
| + | |
| + if debug { | |
| + // Save the cluster for future debugging. | |
| + c.Save(ctx, failureMsg, l) | |
| + | |
| + // Continue with a fresh cluster. | |
| + c = nil | |
| + } else { | |
| + // On any test failure or error, we destroy the cluster. We could be | |
| + // more selective, but this sounds safer. | |
| + l.PrintfCtx(ctx, "destroying cluster %s because: %s", c, failureMsg) | |
| + c.Destroy(context.Background(), closeLogger, l) | |
| + c = nil | |
| } | |
| + | |
| if err != nil { | |
| - // N.B. bail out iff runTest exits exceptionally. | |
| return err | |
| } | |
| } else { | |
| @@ -701,16 +601,21 @@ func allStacks() []byte { | |
| // An error is returned if the test is still running (on another goroutine) when | |
| // this returns. This happens when the test doesn't respond to cancellation. | |
| +// Returns true if the test is considered to have passed, false otherwise. | |
| // | |
| // Args: | |
| // c: The cluster on which the test will run. runTest() does not wipe or destroy | |
| // the cluster. | |
| +// testRunnerLogPath: The path to the test runner's log. It will be copied to | |
| +// the test's artifacts dir if the test fails and we're running under | |
| +// TeamCity. | |
| func (r *testRunner) runTest( | |
| ctx context.Context, | |
| t *testImpl, | |
| runNum int, | |
| runCount int, | |
| c *clusterImpl, | |
| + testRunnerLogPath string, | |
| stdout io.Writer, | |
| l *logger.Logger, | |
| ) error { | |
| @@ -764,6 +669,13 @@ func (r *testRunner) runTest( | |
| if teamCity { | |
| shout(ctx, l, stdout, "##teamcity[testFailed name='%s' details='%s' flowId='%s']", | |
| t.Name(), teamCityEscape(output), runID) | |
| + | |
| + // Copy a snapshot of the testrunner's log to the test's artifacts dir | |
| + // so that we collect it below. | |
| + cp := exec.Command("cp", testRunnerLogPath, t.ArtifactsDir()) | |
| + if err := cp.Run(); err != nil { | |
| + l.ErrorfCtx(ctx, "failed to copy test runner's logs to test artifacts: %s", err) | |
| + } | |
| } | |
| shout(ctx, l, stdout, "--- FAIL: %s (%s)\n%s", runID, durationStr, output) | |
| @@ -835,7 +747,8 @@ func (r *testRunner) runTest( | |
| l.PrintfCtx(ctx, "cluster needs to survive until %s, but has expiration: %s. Extending.", | |
| minExp, c.expiration) | |
| if err := c.Extend(ctx, extend, l); err != nil { | |
| - return errors.Wrapf(err, "failed to extend cluster: %s", c.name) | |
| + t.printfAndFail(0 /* skip */, "failed to extend cluster: %s", err) | |
| + return nil | |
| } | |
| } | |
| @@ -853,6 +766,7 @@ func (r *testRunner) runTest( | |
| go func() { | |
| defer close(testReturnedCh) // closed only after we've grabbed the debug info below | |
| + // This is the call to actually run the test. | |
| defer func() { | |
| // We only have to record panics if the panic'd value is not the sentinel | |
| // produced by t.Fatal*(). | |
| @@ -862,12 +776,10 @@ func (r *testRunner) runTest( | |
| } | |
| }() | |
| - // This is the call to actually run the test. | |
| t.Spec().(*registry.TestSpec).Run(runCtx, t, c) | |
| }() | |
| var timedOut bool | |
| - | |
| select { | |
| case <-testReturnedCh: | |
| s := "success" | |
| @@ -1144,7 +1056,8 @@ func (r *testRunner) generateReport() string { | |
| } | |
| type getWorkCallbacks struct { | |
| - onDestroy func() | |
| + createCluster func(context.Context, testToRunRes) (*clusterImpl, error) | |
| + onDestroy func() | |
| } | |
| // getWork selects the next test to run and creates a suitable cluster for it if | |
| @@ -1154,6 +1067,7 @@ type getWorkCallbacks struct { | |
| // preferred. If a test that can reuse it is not found (or if there's no more | |
| // work), the cluster is destroyed (and so its resources are released). | |
| // | |
| +// If the cluster is to be reused, getWork() wipes it. | |
| func (r *testRunner) getWork( | |
| ctx context.Context, | |
| work *workPool, | |
| @@ -1162,25 +1076,54 @@ func (r *testRunner) getWork( | |
| interrupt <-chan struct{}, | |
| l *logger.Logger, | |
| callbacks getWorkCallbacks, | |
| -) (testToRunRes, error) { | |
| +) (testToRunRes, *clusterImpl, error) { | |
| select { | |
| case <-interrupt: | |
| - return testToRunRes{}, fmt.Errorf("interrupted") | |
| + return testToRunRes{}, nil, fmt.Errorf("interrupted") | |
| default: | |
| } | |
| testToRun, err := work.getTestToRun(ctx, c, qp, r.cr, callbacks.onDestroy, l) | |
| if err != nil { | |
| - return testToRunRes{}, err | |
| + return testToRunRes{}, nil, err | |
| } | |
| if !testToRun.noWork { | |
| l.PrintfCtx(ctx, "Selected test: %s run: %d.", testToRun.spec.Name, testToRun.runNum) | |
| + } | |
| + // Are we done? | |
| + if testToRun.noWork { | |
| + return testToRun, nil, nil | |
| + } | |
| + | |
| + // Create a cluster, if we no longer have one. | |
| + if testToRun.canReuseCluster { | |
| + l.PrintfCtx(ctx, "Using existing cluster: %s. Wiping", c.name) | |
| + if err := c.WipeE(ctx, l); err != nil { | |
| + return testToRunRes{}, nil, err | |
| + } | |
| + if err := c.RunE(ctx, c.All(), "rm -rf "+perfArtifactsDir); err != nil { | |
| + return testToRunRes{}, nil, errors.Wrapf(err, "failed to remove perf artifacts dir") | |
| + } | |
| + if c.localCertsDir != "" { | |
| + if err := os.RemoveAll(c.localCertsDir); err != nil { | |
| + return testToRunRes{}, nil, errors.Wrapf(err, | |
| + "failed to remove local certs in %s", c.localCertsDir) | |
| + } | |
| + c.localCertsDir = "" | |
| + } | |
| + // Overwrite the spec of the cluster with the one coming from the test. In | |
| + // particular, this overwrites the reuse policy to reflect what the test | |
| + // intends to do with it. | |
| + c.spec = testToRun.spec.Cluster | |
| } else { | |
| - // We're done--there are no remaining tests. | |
| - return testToRun, nil | |
| + var err error | |
| + c, err = callbacks.createCluster(ctx, testToRun) | |
| + if err != nil { | |
| + return testToRunRes{}, nil, err | |
| + } | |
| } | |
| - return testToRun, nil | |
| + return testToRun, c, nil | |
| } | |
| // addWorker updates the bookkeeping for one more worker. | |
| diff --git a/pkg/cmd/roachtest/test_test.go b/pkg/cmd/roachtest/test_test.go | |
| index 8d2ebb0b92..b7616af3fc 100644 | |
| --- a/pkg/cmd/roachtest/test_test.go | |
| +++ b/pkg/cmd/roachtest/test_test.go | |
| @@ -26,7 +26,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/cmd/roachtest/test" | |
| "github.com/cockroachdb/cockroach/pkg/roachprod/logger" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| - "github.com/cockroachdb/cockroach/pkg/util/quotapool" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| "github.com/cockroachdb/cockroach/pkg/util/syncutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/version" | |
| @@ -90,16 +89,6 @@ func nilLogger() *logger.Logger { | |
| return l | |
| } | |
| -func alwaysFailingClusterAllocator( | |
| - ctx context.Context, | |
| - t registry.TestSpec, | |
| - alloc *quotapool.IntAlloc, | |
| - artifactsDir string, | |
| - wStatus *workerStatus, | |
| -) (*clusterImpl, error) { | |
| - return nil, errors.New("cluster creation failed") | |
| -} | |
| - | |
| func TestRunnerRun(t *testing.T) { | |
| ctx := context.Background() | |
| r := mkReg(t) | |
| @@ -164,21 +153,11 @@ func TestRunnerRun(t *testing.T) { | |
| cpuQuota: 1000, | |
| keepClustersOnTestFailure: false, | |
| } | |
| - var clusterAllocator clusterAllocatorFn | |
| - // run without cluster allocator error injection | |
| err := runner.Run(ctx, tests, 1, /* count */ | |
| - defaultParallelism, copt, testOpts{}, lopt, clusterAllocator) | |
| - | |
| - assertTestCompletion(t, tests, c.filters, runner.getCompletedTests(), err, c.expErr) | |
| + defaultParallelism, copt, testOpts{}, lopt) | |
| - // N.B. skip the case of no matching tests | |
| - if len(tests) > 0 { | |
| - // run _with_ cluster allocator error injection | |
| - clusterAllocator = alwaysFailingClusterAllocator | |
| - err = runner.Run(ctx, tests, 1, /* count */ | |
| - defaultParallelism, copt, testOpts{}, lopt, clusterAllocator) | |
| - | |
| - assertTestCompletion(t, tests, c.filters, runner.getCompletedTests(), err, "some clusters could not be created") | |
| + if !testutils.IsError(err, c.expErr) { | |
| + t.Fatalf("expected err: %q, but found %v. Filters: %s", c.expErr, err, c.filters) | |
| } | |
| out := stdout.String() + "\n" + stderr.String() | |
| if exp := c.expOut; exp != "" && !strings.Contains(out, exp) { | |
| @@ -188,29 +167,6 @@ func TestRunnerRun(t *testing.T) { | |
| } | |
| } | |
| -// verifies that actual test completion conditions match the expected | |
| -func assertTestCompletion( | |
| - t *testing.T, | |
| - tests []registry.TestSpec, | |
| - filters []string, | |
| - completed []completedTestInfo, | |
| - actualErr error, | |
| - expectedErr string, | |
| -) { | |
| - require.True(t, len(completed) == len(tests)) | |
| - | |
| - for _, info := range completed { | |
| - if info.test == "pass" { | |
| - require.True(t, info.pass) | |
| - } else if info.test == "fail" { | |
| - require.True(t, !info.pass) | |
| - } | |
| - } | |
| - if !testutils.IsError(actualErr, expectedErr) { | |
| - t.Fatalf("expected err: %q, but found %v. Filters: %s", expectedErr, actualErr, filters) | |
| - } | |
| -} | |
| - | |
| type syncedBuffer struct { | |
| mu syncutil.Mutex | |
| buf bytes.Buffer | |
| @@ -259,7 +215,7 @@ func TestRunnerTestTimeout(t *testing.T) { | |
| }, | |
| } | |
| err := runner.Run(ctx, []registry.TestSpec{test}, 1, /* count */ | |
| - defaultParallelism, copt, testOpts{}, lopt, nil /* clusterAllocator */) | |
| + defaultParallelism, copt, testOpts{}, lopt) | |
| if !testutils.IsError(err, "some tests failed") { | |
| t.Fatalf("expected error \"some tests failed\", got: %v", err) | |
| } | |
| @@ -351,7 +307,7 @@ func runExitCodeTest(t *testing.T, injectedError error) error { | |
| stderr: ioutil.Discard, | |
| artifactsDir: "", | |
| } | |
| - return runner.Run(ctx, tests, 1, 1, clustersOpt{}, testOpts{}, lopt, nil /* clusterAllocator */) | |
| + return runner.Run(ctx, tests, 1, 1, clustersOpt{}, testOpts{}, lopt) | |
| } | |
| func TestExitCode(t *testing.T) { | |
| diff --git a/pkg/cmd/roachtest/tests/drain.go b/pkg/cmd/roachtest/tests/drain.go | |
| index 6cf9387f69..7d75784793 100644 | |
| --- a/pkg/cmd/roachtest/tests/drain.go | |
| +++ b/pkg/cmd/roachtest/tests/drain.go | |
| @@ -31,164 +31,166 @@ import ( | |
| func registerDrain(r registry.Registry) { | |
| { | |
| r.Add(registry.TestSpec{ | |
| - Name: "drain/conn-wait", | |
| + Name: "drain/early-exit-conn-wait", | |
| Owner: registry.OwnerSQLExperience, | |
| Cluster: r.MakeClusterSpec(1), | |
| Run: func(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| - runConnectionWait(ctx, t, c) | |
| + runEarlyExitInConnectionWait(ctx, t, c) | |
| }, | |
| - }, | |
| - ) | |
| + }) | |
| + | |
| + r.Add(registry.TestSpec{ | |
| + Name: "drain/warn-conn-wait-timeout", | |
| + Owner: registry.OwnerSQLExperience, | |
| + Cluster: r.MakeClusterSpec(1), | |
| + Run: func(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| + runTestWarningForConnWait(ctx, t, c) | |
| + }, | |
| + }) | |
| } | |
| } | |
| -func runConnectionWait(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| +// runEarlyExitInConnectionWait is to verify that draining proceeds immediately | |
| +// after connections are closed client-side. | |
| +func runEarlyExitInConnectionWait(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| var err error | |
| + const ( | |
| + // Set the duration of each phase of the draining period. | |
| + drainWaitDuration = 5 * time.Second | |
| + connectionWaitDuration = 100 * time.Second | |
| + queryWaitDuration = 10 * time.Second | |
| + // pokeDuringConnWaitTimestamp is the timestamp after the server | |
| + // starts waiting for SQL connections to close (with the start of the whole | |
| + // draining process marked as timestamp 0). It should be set larger than | |
| + // drainWaitDuration, but smaller than (drainWaitDuration + | |
| + // connectionWaitDuration). | |
| + pokeDuringConnWaitTimestamp = 15 * time.Second | |
| + connMaxLifetime = 10 * time.Second | |
| + connMaxCount = 5 | |
| + nodeToDrain = 1 | |
| + ) | |
| + totalWaitDuration := drainWaitDuration + connectionWaitDuration + queryWaitDuration | |
| - err = c.PutE(ctx, t.L(), t.Cockroach(), "./cockroach", c.All()) | |
| - require.NoError(t, err, "cannot mount cockroach binary") | |
| + prepareCluster(ctx, t, c, drainWaitDuration, connectionWaitDuration, queryWaitDuration) | |
| - // Verify that draining proceeds immediately after connections are closed client-side. | |
| - { | |
| - const ( | |
| - // Set the duration of each phase of the draining period. | |
| - drainWaitDuration = 10 * time.Second | |
| - connectionWaitDuration = 100 * time.Second | |
| - queryWaitDuration = 10 * time.Second | |
| - // pokeDuringConnWaitTimestamp is the timestamp after the server | |
| - // starts waiting for SQL connections to close (with the start of the whole | |
| - // draining process marked as timestamp 0). It should be set larger than | |
| - // drainWaitDuration, but smaller than (drainWaitDuration + | |
| - // connectionWaitDuration). | |
| - pokeDuringConnWaitTimestamp = 45 * time.Second | |
| - connMaxLifetime = 30 * time.Second | |
| - connMaxCount = 5 | |
| - nodeToDrain = 1 | |
| - ) | |
| - totalWaitDuration := drainWaitDuration + connectionWaitDuration + queryWaitDuration | |
| - | |
| - prepareCluster(ctx, t, c, drainWaitDuration, connectionWaitDuration, queryWaitDuration) | |
| - | |
| - db := c.Conn(ctx, t.L(), nodeToDrain) | |
| - defer db.Close() | |
| - | |
| - db.SetConnMaxLifetime(connMaxLifetime) | |
| - db.SetMaxOpenConns(connMaxCount) | |
| - | |
| - var conns []*gosql.Conn | |
| - | |
| - // Get two connections from the connection pools. | |
| - for j := 0; j < 2; j++ { | |
| - conn, err := db.Conn(ctx) | |
| - | |
| - require.NoError(t, err, "failed to a SQL connection from the connection pool") | |
| - | |
| - conns = append(conns, conn) | |
| - } | |
| - | |
| - // Start draining the node. | |
| - m := c.NewMonitor(ctx, c.Node(nodeToDrain)) | |
| - | |
| - m.Go(func(ctx context.Context) error { | |
| - t.Status(fmt.Sprintf("start draining node %d", nodeToDrain)) | |
| - return c.RunE(ctx, | |
| - c.Node(nodeToDrain), | |
| - fmt.Sprintf("./cockroach node drain --insecure --drain-wait=%fs", | |
| - totalWaitDuration.Seconds())) | |
| - }) | |
| + db := c.Conn(ctx, t.L(), nodeToDrain) | |
| + defer db.Close() | |
| - drainStartTimestamp := timeutil.Now() | |
| + db.SetConnMaxLifetime(connMaxLifetime) | |
| + db.SetMaxOpenConns(connMaxCount) | |
| - // Sleep till the server is in the status of waiting for users to close SQL | |
| - // connections. Verify that the server is rejecting new SQL connections now. | |
| - time.Sleep(pokeDuringConnWaitTimestamp) | |
| - _, err = db.Conn(ctx) | |
| - if err != nil { | |
| - t.Status(fmt.Sprintf("%s after draining starts, server is rejecting "+ | |
| - "new SQL connections: %v", pokeDuringConnWaitTimestamp, err)) | |
| - } else { | |
| - t.Fatal(errors.New("new SQL connections should not be allowed when the server " + | |
| - "starts waiting for the user to close SQL connections")) | |
| - } | |
| + var conns []*gosql.Conn | |
| - require.Equalf(t, db.Stats().OpenConnections, 2, "number of open connections should be 2") | |
| + // Get two connections from the connection pools. | |
| + for j := 0; j < 2; j++ { | |
| + conn, err := db.Conn(ctx) | |
| - t.Status("number of open connections: ", db.Stats().OpenConnections) | |
| + require.NoError(t, err, "failed to a SQL connection from the connection pool") | |
| - randConn := conns[rand.Intn(len(conns))] | |
| + conns = append(conns, conn) | |
| + } | |
| - // When server is waiting clients to close connections, verify that SQL | |
| - // queries do not fail. | |
| - _, err = randConn.ExecContext(ctx, "SELECT 1;") | |
| + // Start draining the node. | |
| + m := c.NewMonitor(ctx, c.Node(nodeToDrain)) | |
| + | |
| + m.Go(func(ctx context.Context) error { | |
| + t.Status(fmt.Sprintf("start draining node %d", nodeToDrain)) | |
| + return c.RunE(ctx, | |
| + c.Node(nodeToDrain), | |
| + fmt.Sprintf("./cockroach node drain --insecure --drain-wait=%fs", | |
| + totalWaitDuration.Seconds())) | |
| + }) | |
| + | |
| + drainStartTimestamp := timeutil.Now() | |
| + | |
| + // Sleep till the server is in the status of waiting for users to close SQL | |
| + // connections. Verify that the server is rejecting new SQL connections now. | |
| + time.Sleep(pokeDuringConnWaitTimestamp) | |
| + | |
| + if _, err := db.Conn(ctx); err != nil { | |
| + t.Status(fmt.Sprintf("%s after draining starts, server is rejecting "+ | |
| + "new SQL connections: %v", pokeDuringConnWaitTimestamp, err)) | |
| + } else { | |
| + t.Fatal(errors.New("new SQL connections should not be allowed when the server " + | |
| + "starts waiting for the user to close SQL connections")) | |
| + } | |
| - require.NoError(t, err, "expected query not to fail before the "+ | |
| - "server starts draining SQL connections") | |
| + require.Equalf(t, db.Stats().OpenConnections, 2, "number of open connections should be 2") | |
| - for _, conn := range conns { | |
| - err := conn.Close() | |
| - require.NoError(t, err, | |
| - "expected connection to be able to be successfully closed client-side") | |
| - } | |
| + t.Status("number of open connections: ", db.Stats().OpenConnections) | |
| - t.Status("all SQL connections are put back to the connection pool") | |
| + randConn := conns[rand.Intn(len(conns))] | |
| + t.Status("execting sql query with connection %s", randConn) | |
| - err = m.WaitE() | |
| - require.NoError(t, err, "error waiting for the draining to finish") | |
| + // When server is waiting clients to close connections, verify that SQL | |
| + // queries do not fail. | |
| + _, err = randConn.ExecContext(ctx, "SELECT 1;") | |
| + require.NoError(t, err, "expected query not to fail before the "+ | |
| + "server starts draining SQL connections") | |
| - drainEndTimestamp := timeutil.Now() | |
| - actualDrainDuration := drainEndTimestamp.Sub(drainStartTimestamp).Seconds() | |
| + for _, conn := range conns { | |
| + err := conn.Close() | |
| + require.NoError(t, err, | |
| + "expected connection to be able to be successfully closed client-side") | |
| + } | |
| + | |
| + t.Status("all SQL connections are put back to the connection pool") | |
| - t.L().Printf("the draining lasted %f seconds", actualDrainDuration) | |
| + err = m.WaitE() | |
| + require.NoError(t, err, "error waiting for the draining to finish") | |
| - if actualDrainDuration >= float64(totalWaitDuration)-10 { | |
| - t.Fatal(errors.New("the draining process didn't early exit " + | |
| - "when waiting for server to close all SQL connections")) | |
| - } | |
| + drainEndTimestamp := timeutil.Now() | |
| + actualDrainDuration := drainEndTimestamp.Sub(drainStartTimestamp).Seconds() | |
| - // Fully quit the draining node so that we can restart it for the next test. | |
| - quitNode(ctx, t, c, nodeToDrain) | |
| + t.L().Printf("the draining lasted %f seconds", actualDrainDuration) | |
| + | |
| + if actualDrainDuration >= float64(totalWaitDuration)-10 { | |
| + t.Fatal(errors.New("the draining process didn't early exit " + | |
| + "when waiting for server to close all SQL connections")) | |
| } | |
| - // Verify a warning exists in the case that connectionWait expires. | |
| - { | |
| - const ( | |
| - // Set the duration of the draining period. | |
| - drainWaitDuration = 0 * time.Second | |
| - connectionWaitDuration = 10 * time.Second | |
| - queryWaitDuration = 20 * time.Second | |
| - nodeToDrain = 1 | |
| - ) | |
| +} | |
| - totalWaitDuration := drainWaitDuration + connectionWaitDuration + queryWaitDuration | |
| +// runTestWarningForConnWait is to verify a warning exists in the case that | |
| +// connectionWait expires. | |
| +func runTestWarningForConnWait(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| + var err error | |
| + const ( | |
| + // Set the duration of the draining period. | |
| + drainWaitDuration = 0 * time.Second | |
| + connectionWaitDuration = 10 * time.Second | |
| + queryWaitDuration = 20 * time.Second | |
| + nodeToDrain = 1 | |
| + ) | |
| - prepareCluster(ctx, t, c, drainWaitDuration, connectionWaitDuration, queryWaitDuration) | |
| + totalWaitDuration := drainWaitDuration + connectionWaitDuration + queryWaitDuration | |
| - db := c.Conn(ctx, t.L(), nodeToDrain) | |
| - defer db.Close() | |
| + prepareCluster(ctx, t, c, drainWaitDuration, connectionWaitDuration, queryWaitDuration) | |
| - // Get a connection from the connection pool. | |
| - _, err = db.Conn(ctx) | |
| + db := c.Conn(ctx, t.L(), nodeToDrain) | |
| + defer db.Close() | |
| - require.NoError(t, err, "cannot get a SQL connection from the connection pool") | |
| + // Get a connection from the connection pool. | |
| + _, err = db.Conn(ctx) | |
| - m := c.NewMonitor(ctx, c.Node(nodeToDrain)) | |
| - m.Go(func(ctx context.Context) error { | |
| - t.Status(fmt.Sprintf("draining node %d", nodeToDrain)) | |
| - return c.RunE(ctx, | |
| - c.Node(nodeToDrain), | |
| - fmt.Sprintf("./cockroach node drain --insecure --drain-wait=%fs", | |
| - totalWaitDuration.Seconds())) | |
| - }) | |
| + require.NoError(t, err, "cannot get a SQL connection from the connection pool") | |
| - err = m.WaitE() | |
| - require.NoError(t, err, "error waiting for the draining to finish") | |
| + m := c.NewMonitor(ctx, c.Node(nodeToDrain)) | |
| + m.Go(func(ctx context.Context) error { | |
| + t.Status(fmt.Sprintf("draining node %d", nodeToDrain)) | |
| + return c.RunE(ctx, | |
| + c.Node(nodeToDrain), | |
| + fmt.Sprintf("./cockroach node drain --insecure --drain-wait=%fs", | |
| + totalWaitDuration.Seconds())) | |
| + }) | |
| - logFile := filepath.Join("logs", "*.log") | |
| - err = c.RunE(ctx, c.Node(nodeToDrain), | |
| - "grep", "-q", "'proceeding to drain SQL connections'", logFile) | |
| - require.NoError(t, err, "warning is not logged in the log file") | |
| - } | |
| + err = m.WaitE() | |
| + require.NoError(t, err, "error waiting for the draining to finish") | |
| + logFile := filepath.Join("logs", "*.log") | |
| + err = c.RunE(ctx, c.Node(nodeToDrain), | |
| + "grep", "-q", "'proceeding to drain SQL connections'", logFile) | |
| + require.NoError(t, err, "warning is not logged in the log file") | |
| } | |
| // prepareCluster is to start the server on nodes in the given cluster, and set | |
| @@ -201,6 +203,9 @@ func prepareCluster( | |
| connectionWait time.Duration, | |
| queryWait time.Duration, | |
| ) { | |
| + var err error | |
| + err = c.PutE(ctx, t.L(), t.Cockroach(), "./cockroach", c.All()) | |
| + require.NoError(t, err, "cannot mount cockroach binary") | |
| c.Start(ctx, t.L(), option.DefaultStartOpts(), install.MakeClusterSettings(), c.All()) | |
| @@ -215,22 +220,6 @@ func prepareCluster( | |
| connectionWait.Seconds(), | |
| queryWait.Seconds(), | |
| ) | |
| - _, err := db.ExecContext(ctx, waitPhasesSettingStmt) | |
| + _, err = db.ExecContext(ctx, waitPhasesSettingStmt) | |
| require.NoError(t, err, "cannot set cluster setting") | |
| } | |
| - | |
| -func quitNode(ctx context.Context, t test.Test, c cluster.Cluster, node int) { | |
| - args := []string{ | |
| - "./cockroach", "quit", "--insecure", "--logtostderr=INFO", | |
| - fmt.Sprintf("--port={pgport:%d}", node)} | |
| - result, err := c.RunWithDetailsSingleNode(ctx, t.L(), c.Node(node), args...) | |
| - output := result.Stdout + result.Stderr | |
| - t.L().Printf("cockroach quit:\n%s\n", output) | |
| - require.NoError(t, err, "cannot quit cockroach") | |
| - | |
| - stopOpts := option.DefaultStopOpts() | |
| - stopOpts.RoachprodOpts.Sig = 0 | |
| - stopOpts.RoachprodOpts.Wait = true | |
| - c.Stop(ctx, t.L(), stopOpts, c.All()) | |
| - t.L().Printf("stopped cluster") | |
| -} | |
| diff --git a/pkg/cmd/roachtest/tests/drt.go b/pkg/cmd/roachtest/tests/drt.go | |
| index 3841ef186f..f43ad6f26c 100644 | |
| --- a/pkg/cmd/roachtest/tests/drt.go | |
| +++ b/pkg/cmd/roachtest/tests/drt.go | |
| @@ -10,6 +10,8 @@ | |
| package tests | |
| +//go:generate mockgen -package tests -destination drt_generated_test.go . PromClient | |
| + | |
| import ( | |
| "context" | |
| "fmt" | |
| @@ -18,17 +20,21 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/cmd/roachtest/prometheus" | |
| "github.com/cockroachdb/cockroach/pkg/roachprod/logger" | |
| "github.com/cockroachdb/errors" | |
| + promv1 "github.com/prometheus/client_golang/api/prometheus/v1" | |
| "github.com/prometheus/common/model" | |
| ) | |
| -//go:generate mockgen -package tests -destination drt_generated_test.go github.com/cockroachdb/cockroach/pkg/cmd/roachtest/prometheus Client | |
| +// PromClient is an interface allowing queries against Prometheus. | |
| +type PromClient interface { | |
| + Query(ctx context.Context, query string, ts time.Time) (model.Value, promv1.Warnings, error) | |
| +} | |
| type tpccChaosEventProcessor struct { | |
| workloadInstances []workloadInstance | |
| workloadNodeIP string | |
| ops []string | |
| ch chan ChaosEvent | |
| - promClient prometheus.Client | |
| + promClient PromClient | |
| errs []error | |
| // allowZeroSuccessDuringUptime allows 0 successes during an uptime event. | |
| diff --git a/pkg/cmd/roachtest/tests/drt_generated_test.go b/pkg/cmd/roachtest/tests/drt_generated_test.go | |
| index f811a50317..c2ee148e18 100644 | |
| --- a/pkg/cmd/roachtest/tests/drt_generated_test.go | |
| +++ b/pkg/cmd/roachtest/tests/drt_generated_test.go | |
| @@ -1,5 +1,5 @@ | |
| // Code generated by MockGen. DO NOT EDIT. | |
| -// Source: github.com/cockroachdb/cockroach/pkg/cmd/roachtest/prometheus (interfaces: Client) | |
| +// Source: github.com/cockroachdb/cockroach/pkg/cmd/roachtest/tests (interfaces: PromClient) | |
| // Package tests is a generated GoMock package. | |
| package tests | |
| @@ -14,31 +14,31 @@ import ( | |
| model "github.com/prometheus/common/model" | |
| ) | |
| -// MockClient is a mock of Client interface. | |
| -type MockClient struct { | |
| +// MockPromClient is a mock of PromClient interface. | |
| +type MockPromClient struct { | |
| ctrl *gomock.Controller | |
| - recorder *MockClientMockRecorder | |
| + recorder *MockPromClientMockRecorder | |
| } | |
| -// MockClientMockRecorder is the mock recorder for MockClient. | |
| -type MockClientMockRecorder struct { | |
| - mock *MockClient | |
| +// MockPromClientMockRecorder is the mock recorder for MockPromClient. | |
| +type MockPromClientMockRecorder struct { | |
| + mock *MockPromClient | |
| } | |
| -// NewMockClient creates a new mock instance. | |
| -func NewMockClient(ctrl *gomock.Controller) *MockClient { | |
| - mock := &MockClient{ctrl: ctrl} | |
| - mock.recorder = &MockClientMockRecorder{mock} | |
| +// NewMockPromClient creates a new mock instance. | |
| +func NewMockPromClient(ctrl *gomock.Controller) *MockPromClient { | |
| + mock := &MockPromClient{ctrl: ctrl} | |
| + mock.recorder = &MockPromClientMockRecorder{mock} | |
| return mock | |
| } | |
| // EXPECT returns an object that allows the caller to indicate expected use. | |
| -func (m *MockClient) EXPECT() *MockClientMockRecorder { | |
| +func (m *MockPromClient) EXPECT() *MockPromClientMockRecorder { | |
| return m.recorder | |
| } | |
| // Query mocks base method. | |
| -func (m *MockClient) Query(arg0 context.Context, arg1 string, arg2 time.Time) (model.Value, v1.Warnings, error) { | |
| +func (m *MockPromClient) Query(arg0 context.Context, arg1 string, arg2 time.Time) (model.Value, v1.Warnings, error) { | |
| m.ctrl.T.Helper() | |
| ret := m.ctrl.Call(m, "Query", arg0, arg1, arg2) | |
| ret0, _ := ret[0].(model.Value) | |
| @@ -48,7 +48,7 @@ func (m *MockClient) Query(arg0 context.Context, arg1 string, arg2 time.Time) (m | |
| } | |
| // Query indicates an expected call of Query. | |
| -func (mr *MockClientMockRecorder) Query(arg0, arg1, arg2 interface{}) *gomock.Call { | |
| +func (mr *MockPromClientMockRecorder) Query(arg0, arg1, arg2 interface{}) *gomock.Call { | |
| mr.mock.ctrl.T.Helper() | |
| - return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "Query", reflect.TypeOf((*MockClient)(nil).Query), arg0, arg1, arg2) | |
| + return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "Query", reflect.TypeOf((*MockPromClient)(nil).Query), arg0, arg1, arg2) | |
| } | |
| diff --git a/pkg/cmd/roachtest/tests/drt_test.go b/pkg/cmd/roachtest/tests/drt_test.go | |
| index 7715a1f8f8..01a52bbcf8 100644 | |
| --- a/pkg/cmd/roachtest/tests/drt_test.go | |
| +++ b/pkg/cmd/roachtest/tests/drt_test.go | |
| @@ -518,8 +518,8 @@ func TestTPCCChaosEventProcessor(t *testing.T) { | |
| allowZeroSuccessDuringUptime: tc.allowZeroSuccessDuringUptime, | |
| maxErrorsDuringUptime: tc.maxErrorsDuringUptime, | |
| - promClient: func(ctrl *gomock.Controller) prometheus.Client { | |
| - c := NewMockClient(ctrl) | |
| + promClient: func(ctrl *gomock.Controller) PromClient { | |
| + c := NewMockPromClient(ctrl) | |
| e := c.EXPECT() | |
| for _, m := range tc.mockPromQueries { | |
| e.Query(ctx, m.q, m.t).Return( | |
| diff --git a/pkg/cmd/roachtest/tests/slow_drain.go b/pkg/cmd/roachtest/tests/slow_drain.go | |
| index a8c97b760d..2d4f3a5125 100644 | |
| --- a/pkg/cmd/roachtest/tests/slow_drain.go | |
| +++ b/pkg/cmd/roachtest/tests/slow_drain.go | |
| @@ -27,7 +27,7 @@ import ( | |
| func registerSlowDrain(r registry.Registry) { | |
| numNodes := 6 | |
| - duration := time.Minute | |
| + duration := 30 * time.Minute | |
| r.Add(registry.TestSpec{ | |
| Name: fmt.Sprintf("slow-drain/duration=%s", duration), | |
| @@ -55,6 +55,7 @@ func runSlowDrain(ctx context.Context, t test.Test, c cluster.Cluster, duration | |
| require.NoError(t, err) | |
| c.Start(ctx, t.L(), option.DefaultStartOpts(), install.MakeClusterSettings(), c.All()) | |
| + c.Run(ctx, c.Node(pinnedNodeID), `./cockroach workload init kv --drop --splits 1000`) | |
| run := func(stmt string) { | |
| db := c.Conn(ctx, t.L(), pinnedNodeID) | |
| @@ -86,30 +87,34 @@ func runSlowDrain(ctx context.Context, t test.Test, c cluster.Cluster, duration | |
| } | |
| } | |
| - { | |
| - db := c.Conn(ctx, t.L(), pinnedNodeID) | |
| - defer db.Close() | |
| + run(fmt.Sprintf(`ALTER RANGE default CONFIGURE ZONE USING num_replicas=%d`, replicationFactor)) | |
| + run(fmt.Sprintf(`ALTER DATABASE system CONFIGURE ZONE USING num_replicas=%d`, replicationFactor)) | |
| - // Set the replication factor. | |
| - run(fmt.Sprintf(`ALTER RANGE default CONFIGURE ZONE USING num_replicas=%d`, replicationFactor)) | |
| - run(fmt.Sprintf(`ALTER DATABASE system CONFIGURE ZONE USING num_replicas=%d`, replicationFactor)) | |
| + db := c.Conn(ctx, t.L(), pinnedNodeID) | |
| + defer db.Close() | |
| - // Wait for initial up-replication. | |
| - waitForReplication(db) | |
| - } | |
| + waitForReplication(db) | |
| - // Drain the last 5 nodes from the cluster, resulting in immovable leases on | |
| - // at least one of the nodes. | |
| m := c.NewMonitor(ctx) | |
| + m.Go(func(ctx context.Context) error { | |
| + return c.RunE(ctx, c.Node(pinnedNodeID), | |
| + fmt.Sprintf("./cockroach workload run kv --max-rate 500 --tolerate-errors --duration=%s {pgurl:1-6}", | |
| + duration.String(), | |
| + ), | |
| + ) | |
| + }, | |
| + ) | |
| + | |
| + // Let the workload run for a small amount of time. | |
| + time.Sleep(1 * time.Minute) | |
| + | |
| + // Drain the last 5 nodes from the cluster, resulting in immovable leases on at least one of the nodes. | |
| for nodeID := 2; nodeID <= numNodes; nodeID++ { | |
| id := nodeID | |
| m.Go(func(ctx context.Context) error { | |
| drain := func(id int) error { | |
| t.Status(fmt.Sprintf("draining node %d", id)) | |
| - return c.RunE(ctx, | |
| - c.Node(id), | |
| - fmt.Sprintf("./cockroach node drain %d --insecure --drain-wait=%s", id, duration.String()), | |
| - ) | |
| + return c.RunE(ctx, c.Node(id), fmt.Sprintf("./cockroach node drain %d --insecure", id)) | |
| } | |
| return drain(id) | |
| }) | |
| @@ -120,7 +125,6 @@ func runSlowDrain(ctx context.Context, t test.Test, c cluster.Cluster, duration | |
| // Check for more verbose logging concerning lease transfer stalls. | |
| // The extra logging should exist on the logs of at least one of the nodes. | |
| - t.Status("checking for stalling drain logging...") | |
| found := false | |
| for nodeID := 2; nodeID <= numNodes; nodeID++ { | |
| if err := c.RunE(ctx, c.Node(nodeID), | |
| @@ -130,10 +134,8 @@ func runSlowDrain(ctx context.Context, t test.Test, c cluster.Cluster, duration | |
| } | |
| } | |
| require.True(t, found) | |
| - t.Status("log messages found") | |
| - // Expect the drain timeout to expire. | |
| - t.Status("waiting for the drain timeout to elapse...") | |
| + // Expect a failed drain. | |
| err = m.WaitE() | |
| require.Error(t, err) | |
| } | |
| diff --git a/pkg/cmd/roachtest/tests/sstable_corruption.go b/pkg/cmd/roachtest/tests/sstable_corruption.go | |
| index 237591ba54..d90b315d21 100644 | |
| --- a/pkg/cmd/roachtest/tests/sstable_corruption.go | |
| +++ b/pkg/cmd/roachtest/tests/sstable_corruption.go | |
| @@ -43,7 +43,7 @@ func runSSTableCorruption(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| // to have multiple ranges, and some sstables with only table keys. | |
| t.Status("importing tpcc fixture") | |
| c.Run(ctx, workloadNode, | |
| - "./cockroach workload fixtures import tpcc --warehouses=500 --fks=false --checks=false") | |
| + "./cockroach workload fixtures import tpcc --warehouses=100 --fks=false --checks=false") | |
| return nil | |
| }) | |
| m.Wait() | |
| @@ -55,7 +55,7 @@ func runSSTableCorruption(ctx context.Context, t test.Test, c cluster.Cluster) { | |
| result, err := c.RunWithDetailsSingleNode(ctx, t.L(), c.Node(node), | |
| "./cockroach debug pebble manifest dump {store-dir}/MANIFEST-* | grep -v added | grep -v deleted | grep \"\\[/Table\"") | |
| if err != nil { | |
| - t.Fatalf("could not find tables to corrupt: %s\nstdout: %s\nstderr: %s", err, result.Stdout, result.Stderr) | |
| + t.Fatal(err) | |
| } | |
| tableSSTs := strings.Split(result.Stdout, "\n") | |
| if len(tableSSTs) == 0 { | |
| diff --git a/pkg/cmd/roachtest/work_pool.go b/pkg/cmd/roachtest/work_pool.go | |
| index 30ea8e7d70..2475ff9341 100644 | |
| --- a/pkg/cmd/roachtest/work_pool.go | |
| +++ b/pkg/cmd/roachtest/work_pool.go | |
| @@ -172,9 +172,6 @@ func (p *workPool) selectTestForCluster( | |
| // If multiple tests are eligible to run, one with the most runs left is chosen. | |
| // TODO(andrei): We could be smarter in guessing what kind of cluster is best to | |
| // allocate. | |
| -// | |
| -// ensures: !testToRunRes.noWork || error == nil | |
| -// | |
| func (p *workPool) selectTest(ctx context.Context, qp *quotapool.IntPool) (testToRunRes, error) { | |
| var ttr testToRunRes | |
| alloc, err := qp.AcquireFunc(ctx, func(ctx context.Context, pi quotapool.PoolInfo) (uint64, error) { | |
| diff --git a/pkg/gen/docs.bzl b/pkg/gen/docs.bzl | |
| index 73faa4a4d0..7af9c489b3 100644 | |
| --- a/pkg/gen/docs.bzl | |
| +++ b/pkg/gen/docs.bzl | |
| @@ -50,7 +50,6 @@ DOCS_SRCS = [ | |
| "//docs/generated/sql/bnf:alter_table_partition_by.bnf", | |
| "//docs/generated/sql/bnf:alter_table_set_schema_stmt.bnf", | |
| "//docs/generated/sql/bnf:alter_table_stmt.bnf", | |
| - "//docs/generated/sql/bnf:alter_tenant_csetting_stmt.bnf", | |
| "//docs/generated/sql/bnf:alter_type.bnf", | |
| "//docs/generated/sql/bnf:alter_view.bnf", | |
| "//docs/generated/sql/bnf:alter_view_owner_stmt.bnf", | |
| @@ -184,7 +183,6 @@ DOCS_SRCS = [ | |
| "//docs/generated/sql/bnf:set_exprs_internal.bnf", | |
| "//docs/generated/sql/bnf:set_local_stmt.bnf", | |
| "//docs/generated/sql/bnf:set_operation.bnf", | |
| - "//docs/generated/sql/bnf:set_or_reset_csetting_stmt.bnf", | |
| "//docs/generated/sql/bnf:set_rest.bnf", | |
| "//docs/generated/sql/bnf:set_rest_more.bnf", | |
| "//docs/generated/sql/bnf:set_session_stmt.bnf", | |
| @@ -204,7 +202,6 @@ DOCS_SRCS = [ | |
| "//docs/generated/sql/bnf:show_indexes_stmt.bnf", | |
| "//docs/generated/sql/bnf:show_jobs.bnf", | |
| "//docs/generated/sql/bnf:show_keys.bnf", | |
| - "//docs/generated/sql/bnf:show_local_or_tenant_csettings_stmt.bnf", | |
| "//docs/generated/sql/bnf:show_locality.bnf", | |
| "//docs/generated/sql/bnf:show_locality_stmt.bnf", | |
| "//docs/generated/sql/bnf:show_partitions_stmt.bnf", | |
| diff --git a/pkg/gen/protobuf.bzl b/pkg/gen/protobuf.bzl | |
| index f931812340..df87c11dff 100644 | |
| --- a/pkg/gen/protobuf.bzl | |
| +++ b/pkg/gen/protobuf.bzl | |
| @@ -20,7 +20,6 @@ PROTOBUF_SRCS = [ | |
| "//pkg/kv/kvnemesis:kvnemesis_go_proto", | |
| "//pkg/kv/kvserver/closedts/ctpb:ctpb_go_proto", | |
| "//pkg/kv/kvserver/concurrency/lock:lock_go_proto", | |
| - "//pkg/kv/kvserver/concurrency/poison:poison_go_proto", | |
| "//pkg/kv/kvserver/kvserverpb:kvserverpb_go_proto", | |
| "//pkg/kv/kvserver/liveness/livenesspb:livenesspb_go_proto", | |
| "//pkg/kv/kvserver/loqrecovery/loqrecoverypb:loqrecoverypb_go_proto", | |
| @@ -39,7 +38,6 @@ PROTOBUF_SRCS = [ | |
| "//pkg/sql/contentionpb:contentionpb_go_proto", | |
| "//pkg/sql/execinfrapb:execinfrapb_go_proto", | |
| "//pkg/sql/inverted:inverted_go_proto", | |
| - "//pkg/sql/lex:lex_go_proto", | |
| "//pkg/sql/pgwire/pgerror:pgerror_go_proto", | |
| "//pkg/sql/protoreflect/test:protoreflecttest_go_proto", | |
| "//pkg/sql/rowenc/rowencpb:rowencpb_go_proto", | |
| diff --git a/pkg/gossip/infostore.go b/pkg/gossip/infostore.go | |
| index 0933a73ec1..87c2a1223a 100644 | |
| --- a/pkg/gossip/infostore.go | |
| +++ b/pkg/gossip/infostore.go | |
| @@ -28,7 +28,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/syncutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| ) | |
| type stringMatcher interface { | |
| @@ -254,7 +253,7 @@ func (is *infoStore) addInfo(key string, i *Info) error { | |
| if highWaterStamp, ok := is.highWaterStamps[i.NodeID]; ok && highWaterStamp >= i.OrigStamp { | |
| // Report both timestamps in the crash. | |
| log.Fatalf(context.Background(), | |
| - "high water stamp %d >= %d", redact.Safe(highWaterStamp), redact.Safe(i.OrigStamp)) | |
| + "high water stamp %d >= %d", log.Safe(highWaterStamp), log.Safe(i.OrigStamp)) | |
| } | |
| } | |
| // Update info map. | |
| diff --git a/pkg/jobs/BUILD.bazel b/pkg/jobs/BUILD.bazel | |
| index d25942ec95..040ae7af44 100644 | |
| --- a/pkg/jobs/BUILD.bazel | |
| +++ b/pkg/jobs/BUILD.bazel | |
| @@ -89,7 +89,6 @@ go_test( | |
| "testutils_test.go", | |
| ], | |
| embed = [":jobs"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/clusterversion", | |
| diff --git a/pkg/jobs/job_scheduler.go b/pkg/jobs/job_scheduler.go | |
| index b6ed842f46..e0a1390af6 100644 | |
| --- a/pkg/jobs/job_scheduler.go | |
| +++ b/pkg/jobs/job_scheduler.go | |
| @@ -30,7 +30,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/metric" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| - "github.com/cockroachdb/cockroach/pkg/util/syncutil" | |
| "github.com/cockroachdb/errors" | |
| "github.com/cockroachdb/logtags" | |
| ) | |
| @@ -376,54 +375,6 @@ func (s *jobScheduler) schedulerEnabledOnThisNode(ctx context.Context) bool { | |
| return enabled | |
| } | |
| -type syncCancelFunc struct { | |
| - syncutil.Mutex | |
| - context.CancelFunc | |
| -} | |
| - | |
| -// newCancelWhenDisabled arranges for scheduler enabled setting callback to cancel | |
| -// currently executing context. | |
| -func newCancelWhenDisabled(sv *settings.Values) *syncCancelFunc { | |
| - sf := &syncCancelFunc{} | |
| - schedulerEnabledSetting.SetOnChange(sv, func(ctx context.Context) { | |
| - if !schedulerEnabledSetting.Get(sv) { | |
| - sf.Lock() | |
| - if sf.CancelFunc != nil { | |
| - sf.CancelFunc() | |
| - } | |
| - sf.Unlock() | |
| - } | |
| - }) | |
| - return sf | |
| -} | |
| - | |
| -// withCancelOnDisabled executes provided function with the context which will be cancelled | |
| -// if scheduler is disabled. | |
| -func (sf *syncCancelFunc) withCancelOnDisabled( | |
| - ctx context.Context, sv *settings.Values, f func(ctx context.Context) error, | |
| -) error { | |
| - ctx, cancel := func() (context.Context, context.CancelFunc) { | |
| - sf.Lock() | |
| - defer sf.Unlock() | |
| - | |
| - ctx, cancel := context.WithCancel(ctx) | |
| - sf.CancelFunc = cancel | |
| - | |
| - if !schedulerEnabledSetting.Get(sv) { | |
| - cancel() | |
| - } | |
| - | |
| - return ctx, func() { | |
| - sf.Lock() | |
| - defer sf.Unlock() | |
| - cancel() | |
| - sf.CancelFunc = nil | |
| - } | |
| - }() | |
| - defer cancel() | |
| - return f(ctx) | |
| -} | |
| - | |
| func (s *jobScheduler) runDaemon(ctx context.Context, stopper *stop.Stopper) { | |
| _ = stopper.RunAsyncTask(ctx, "job-scheduler", func(ctx context.Context) { | |
| initialDelay := getInitialScanDelay(s.TestingKnobs) | |
| @@ -433,8 +384,6 @@ func (s *jobScheduler) runDaemon(ctx context.Context, stopper *stop.Stopper) { | |
| log.Errorf(ctx, "error registering executor metrics: %+v", err) | |
| } | |
| - whenDisabled := newCancelWhenDisabled(&s.Settings.SV) | |
| - | |
| for timer := time.NewTimer(initialDelay); ; timer.Reset( | |
| getWaitPeriod(ctx, &s.Settings.SV, s.schedulerEnabledOnThisNode, jitter, s.TestingKnobs)) { | |
| select { | |
| @@ -446,11 +395,10 @@ func (s *jobScheduler) runDaemon(ctx context.Context, stopper *stop.Stopper) { | |
| } | |
| maxSchedules := schedulerMaxJobsPerIterationSetting.Get(&s.Settings.SV) | |
| - if err := whenDisabled.withCancelOnDisabled(ctx, &s.Settings.SV, func(ctx context.Context) error { | |
| - return s.DB.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| - return s.executeSchedules(ctx, maxSchedules, txn) | |
| - }) | |
| - }); err != nil { | |
| + err := s.DB.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| + return s.executeSchedules(ctx, maxSchedules, txn) | |
| + }) | |
| + if err != nil { | |
| log.Errorf(ctx, "error executing schedules: %+v", err) | |
| } | |
| diff --git a/pkg/jobs/job_scheduler_test.go b/pkg/jobs/job_scheduler_test.go | |
| index ea9e390f9b..4a2588f183 100644 | |
| --- a/pkg/jobs/job_scheduler_test.go | |
| +++ b/pkg/jobs/job_scheduler_test.go | |
| @@ -826,84 +826,3 @@ func TestSchedulerCanBeRestrictedToSingleNode(t *testing.T) { | |
| }) | |
| } | |
| } | |
| - | |
| -type blockUntilCancelledExecutor struct { | |
| - started, done chan struct{} | |
| -} | |
| - | |
| -var _ ScheduledJobExecutor = (*blockUntilCancelledExecutor)(nil) | |
| - | |
| -func (e *blockUntilCancelledExecutor) ExecuteJob( | |
| - ctx context.Context, | |
| - cfg *scheduledjobs.JobExecutionConfig, | |
| - env scheduledjobs.JobSchedulerEnv, | |
| - schedule *ScheduledJob, | |
| - txn *kv.Txn, | |
| -) error { | |
| - defer close(e.done) | |
| - close(e.started) | |
| - <-ctx.Done() | |
| - return ctx.Err() | |
| -} | |
| - | |
| -func (e *blockUntilCancelledExecutor) NotifyJobTermination( | |
| - ctx context.Context, | |
| - jobID jobspb.JobID, | |
| - jobStatus Status, | |
| - details jobspb.Details, | |
| - env scheduledjobs.JobSchedulerEnv, | |
| - schedule *ScheduledJob, | |
| - ex sqlutil.InternalExecutor, | |
| - txn *kv.Txn, | |
| -) error { | |
| - return nil | |
| -} | |
| - | |
| -func (e *blockUntilCancelledExecutor) Metrics() metric.Struct { | |
| - return nil | |
| -} | |
| - | |
| -func (e *blockUntilCancelledExecutor) GetCreateScheduleStatement( | |
| - ctx context.Context, | |
| - env scheduledjobs.JobSchedulerEnv, | |
| - txn *kv.Txn, | |
| - sj *ScheduledJob, | |
| - ex sqlutil.InternalExecutor, | |
| -) (string, error) { | |
| - return "", errors.AssertionFailedf("unexpected GetCreateScheduleStatement call") | |
| -} | |
| - | |
| -func TestDisablingSchedulerCancelsSchedules(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - const executorName = "block-until-cancelled-executor" | |
| - ex := &blockUntilCancelledExecutor{ | |
| - started: make(chan struct{}), | |
| - done: make(chan struct{}), | |
| - } | |
| - defer registerScopedScheduledJobExecutor(executorName, ex)() | |
| - | |
| - knobs := base.TestingKnobs{ | |
| - JobsTestingKnobs: fastDaemonKnobs(overridePaceSetting(10 * time.Millisecond)), | |
| - } | |
| - ts, _, _ := serverutils.StartServer(t, base.TestServerArgs{Knobs: knobs}) | |
| - defer ts.Stopper().Stop(context.Background()) | |
| - | |
| - // Create schedule which blocks until its context cancelled due to disabled scheduler. | |
| - // We only need to create one schedule. This is because | |
| - // scheduler executes its batch of schedules sequentially, and so, creating more | |
| - // than one doesn't change anything since we block. | |
| - schedule := NewScheduledJob(scheduledjobs.ProdJobSchedulerEnv) | |
| - schedule.SetScheduleLabel("test schedule") | |
| - schedule.SetOwner(security.TestUserName()) | |
| - schedule.SetNextRun(timeutil.Now()) | |
| - schedule.SetExecutionDetails(executorName, jobspb.ExecutionArguments{}) | |
| - require.NoError(t, schedule.Create( | |
| - context.Background(), ts.InternalExecutor().(sqlutil.InternalExecutor), nil)) | |
| - | |
| - <-ex.started | |
| - // Disable scheduler and verify all running schedules were cancelled. | |
| - schedulerEnabledSetting.Override(context.Background(), &ts.ClusterSettings().SV, false) | |
| - <-ex.done | |
| -} | |
| diff --git a/pkg/keysbase/BUILD.bazel b/pkg/keysbase/BUILD.bazel | |
| deleted file mode 100644 | |
| index 8577d05c14..0000000000 | |
| --- a/pkg/keysbase/BUILD.bazel | |
| +++ /dev/null | |
| @@ -1,8 +0,0 @@ | |
| -load("@io_bazel_rules_go//go:def.bzl", "go_library") | |
| - | |
| -go_library( | |
| - name = "keysbase", | |
| - srcs = ["data.go"], | |
| - importpath = "github.com/cockroachdb/cockroach/pkg/keysbase", | |
| - visibility = ["//visibility:public"], | |
| -) | |
| diff --git a/pkg/keysbase/data.go b/pkg/keysbase/data.go | |
| deleted file mode 100644 | |
| index f7f2de9bee..0000000000 | |
| --- a/pkg/keysbase/data.go | |
| +++ /dev/null | |
| @@ -1,36 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package keysbase | |
| - | |
| -// KeyMax is a maximum key value which sorts after all other keys. | |
| -var KeyMax = []byte{0xff, 0xff} | |
| - | |
| -// PrefixEnd determines the end key given b as a prefix, that is the key that | |
| -// sorts precisely behind all keys starting with prefix: "1" is added to the | |
| -// final byte and the carry propagated. The special cases of nil and KeyMin | |
| -// always returns KeyMax. | |
| -func PrefixEnd(b []byte) []byte { | |
| - if len(b) == 0 { | |
| - return KeyMax | |
| - } | |
| - // Switched to "make and copy" pattern in #4963 for performance. | |
| - end := make([]byte, len(b)) | |
| - copy(end, b) | |
| - for i := len(end) - 1; i >= 0; i-- { | |
| - end[i] = end[i] + 1 | |
| - if end[i] != 0 { | |
| - return end[:i+1] | |
| - } | |
| - } | |
| - // This statement will only be reached if the key is already a maximal byte | |
| - // string (i.e. already \xff...). | |
| - return b | |
| -} | |
| diff --git a/pkg/kv/BUILD.bazel b/pkg/kv/BUILD.bazel | |
| index 12eaaf6ffa..337b13521f 100644 | |
| --- a/pkg/kv/BUILD.bazel | |
| +++ b/pkg/kv/BUILD.bazel | |
| @@ -56,7 +56,6 @@ go_test( | |
| ], | |
| data = glob(["testdata/**"]), | |
| embed = [":kv"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/config/zonepb", | |
| diff --git a/pkg/kv/bulk/BUILD.bazel b/pkg/kv/bulk/BUILD.bazel | |
| index 3d7fcdf1a8..77310783c1 100644 | |
| --- a/pkg/kv/bulk/BUILD.bazel | |
| +++ b/pkg/kv/bulk/BUILD.bazel | |
| @@ -13,7 +13,6 @@ go_library( | |
| visibility = ["//visibility:public"], | |
| deps = [ | |
| "//pkg/keys", | |
| - "//pkg/kv", | |
| "//pkg/kv/kvclient/rangecache", | |
| "//pkg/kv/kvserver", | |
| "//pkg/kv/kvserver/kvserverbase", | |
| @@ -29,6 +28,7 @@ go_library( | |
| "//pkg/util/mon", | |
| "//pkg/util/timeutil", | |
| "@com_github_cockroachdb_errors//:errors", | |
| + "@com_github_cockroachdb_redact//:redact", | |
| ], | |
| ) | |
| diff --git a/pkg/kv/bulk/buffering_adder.go b/pkg/kv/bulk/buffering_adder.go | |
| index 28f2f4a52e..59e9484839 100644 | |
| --- a/pkg/kv/bulk/buffering_adder.go | |
| +++ b/pkg/kv/bulk/buffering_adder.go | |
| @@ -16,8 +16,6 @@ import ( | |
| "strings" | |
| "time" | |
| - "github.com/cockroachdb/cockroach/pkg/keys" | |
| - "github.com/cockroachdb/cockroach/pkg/kv" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvclient/rangecache" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverbase" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -49,19 +47,16 @@ type BufferingAdder struct { | |
| // currently buffered kvs. | |
| curBuf kvBuf | |
| - kvSize sz | |
| sorted bool | |
| initialSplits int | |
| - lastFlush time.Time | |
| flushCounts struct { | |
| - total int | |
| - bufferSize int | |
| - totalSort time.Duration | |
| - totalFlush time.Duration | |
| - totalFilling time.Duration | |
| + total int | |
| + bufferSize int | |
| + totalSort time.Duration | |
| + totalFlush time.Duration | |
| } | |
| // name of the BufferingAdder for the purpose of logging only. | |
| @@ -81,7 +76,7 @@ var _ kvserverbase.BulkAdder = &BufferingAdder{} | |
| // encounter an error and need to be split and retired to be applied. | |
| func MakeBulkAdder( | |
| ctx context.Context, | |
| - db *kv.DB, | |
| + db SSTSender, | |
| rangeCache *rangecache.RangeCache, | |
| settings *cluster.Settings, | |
| timestamp hlc.Timestamp, | |
| @@ -97,15 +92,27 @@ func MakeBulkAdder( | |
| if opts.StepBufferSize == 0 { | |
| opts.StepBufferSize = 32 << 20 | |
| } | |
| + if opts.SSTSize == nil { | |
| + opts.SSTSize = func() int64 { return 16 << 20 } | |
| + } | |
| + if opts.SplitAndScatterAfter == nil { | |
| + // splitting _before_ hitting max reduces chance of auto-splitting after the | |
| + // range is full and is more expensive to split/move. | |
| + opts.SplitAndScatterAfter = func() int64 { return 48 << 20 } | |
| + } else if opts.SplitAndScatterAfter() == kvserverbase.DisableExplicitSplits { | |
| + opts.SplitAndScatterAfter = nil | |
| + } | |
| b := &BufferingAdder{ | |
| name: opts.Name, | |
| sink: SSTBatcher{ | |
| db: db, | |
| + maxSize: opts.SSTSize, | |
| rc: rangeCache, | |
| settings: settings, | |
| skipDuplicates: opts.SkipDuplicates, | |
| disallowShadowingBelow: opts.DisallowShadowingBelow, | |
| + splitAfter: opts.SplitAndScatterAfter, | |
| batchTS: opts.BatchTimestamp, | |
| writeAtBatchTS: opts.WriteAtBatchTimestamp, | |
| }, | |
| @@ -116,7 +123,6 @@ func MakeBulkAdder( | |
| bulkMon: bulkMon, | |
| sorted: true, | |
| initialSplits: opts.InitialSplitsIfUnordered, | |
| - lastFlush: timeutil.Now(), | |
| } | |
| // If no monitor is attached to the instance of a bulk adder, we do not | |
| @@ -148,29 +154,14 @@ func (b *BufferingAdder) SetOnFlush(fn func(summary roachpb.BulkOpSummary)) { | |
| // Close closes the underlying SST builder. | |
| func (b *BufferingAdder) Close(ctx context.Context) { | |
| - log.VEventf(ctx, 1, | |
| - "%s adder ingested %s (%s); spent %s filling, %v sorting, %v flushing (%v sink, %v sending, %v splitting, %v scattering %v)", | |
| + log.VEventf(ctx, 2, | |
| + "bulk adder %s ingested %s, flushed %d due to buffer (%s) size. Flushed chunked as %d files (%d after split-retries), %d due to ranges, %d due to sst size.", | |
| b.name, | |
| sz(b.sink.totalRows.DataSize), | |
| - sorted(b.sorted), | |
| - timing(b.flushCounts.totalFilling), | |
| - timing(b.flushCounts.totalSort), | |
| - timing(b.flushCounts.totalFlush), | |
| - timing(b.sink.flushCounts.flushWait), | |
| - timing(b.sink.flushCounts.sendWait), | |
| - timing(b.sink.flushCounts.splitWait), | |
| - timing(b.sink.flushCounts.scatterWait), | |
| - b.sink.flushCounts.scatterMoved, | |
| - ) | |
| - log.VEventf(ctx, 2, "%s adder flushed %d times, %d due to buffer size (%s); flushing chunked into %d files (%d for ranges, %d for sst size, +%d after split-retries)", | |
| - b.name, | |
| - b.flushCounts.total, | |
| b.flushCounts.bufferSize, | |
| sz(b.memAcc.Used()), | |
| - b.sink.flushCounts.total, | |
| - b.sink.flushCounts.split, | |
| - b.sink.flushCounts.sstSize, | |
| - b.sink.flushCounts.files-b.sink.flushCounts.total, | |
| + b.sink.flushCounts.total, b.sink.flushCounts.files, | |
| + b.sink.flushCounts.split, b.sink.flushCounts.sstSize, | |
| ) | |
| b.sink.Close() | |
| @@ -190,9 +181,8 @@ func (b *BufferingAdder) Add(ctx context.Context, key roachpb.Key, value []byte) | |
| if err := b.curBuf.append(key, value); err != nil { | |
| return err | |
| } | |
| - b.kvSize += sz(len(key) + len(value)) | |
| - if b.curBuf.MemSize > sz(b.curBufferSize) { | |
| + if b.curBuf.MemSize > int(b.curBufferSize) { | |
| // This is an optimization to try and increase the current buffer size if | |
| // our memory account permits it. This would lead to creation of a fewer | |
| // number of SSTs. | |
| @@ -213,14 +203,6 @@ func (b *BufferingAdder) Add(ctx context.Context, key roachpb.Key, value []byte) | |
| return nil | |
| } | |
| -func (b *BufferingAdder) bufferedKeys() int { | |
| - return len(b.curBuf.entries) | |
| -} | |
| - | |
| -func (b *BufferingAdder) bufferedMemSize() sz { | |
| - return b.curBuf.MemSize | |
| -} | |
| - | |
| // CurrentBufferFill returns the current buffer fill percentage. | |
| func (b *BufferingAdder) CurrentBufferFill() float32 { | |
| return float32(b.curBuf.MemSize) / float32(b.curBufferSize) | |
| @@ -233,8 +215,7 @@ func (b *BufferingAdder) IsEmpty() bool { | |
| func (b *BufferingAdder) sizeFlush(ctx context.Context) error { | |
| b.flushCounts.bufferSize++ | |
| - log.VEventf(ctx, 2, "%s adder triggering flush of %s of KVs in %s buffer", | |
| - b.name, b.kvSize, b.bufferedMemSize()) | |
| + log.VEventf(ctx, 3, "buffer size triggering flush of %s buffer", sz(b.curBuf.MemSize)) | |
| return b.doFlush(ctx, true) | |
| } | |
| @@ -244,13 +225,10 @@ func (b *BufferingAdder) Flush(ctx context.Context) error { | |
| } | |
| func (b *BufferingAdder) doFlush(ctx context.Context, forSize bool) error { | |
| - b.flushCounts.totalFilling += timeutil.Since(b.lastFlush) | |
| - | |
| - if b.bufferedKeys() == 0 { | |
| + if b.curBuf.Len() == 0 { | |
| if b.onFlush != nil { | |
| b.onFlush(b.sink.GetBatchSummary()) | |
| } | |
| - b.lastFlush = timeutil.Now() | |
| return nil | |
| } | |
| if err := b.sink.Reset(ctx); err != nil { | |
| @@ -273,14 +251,10 @@ func (b *BufferingAdder) doFlush(ctx context.Context, forSize bool) error { | |
| // If this is the first flush and is due to size, if it was unsorted then | |
| // create initial splits if requested before flushing. | |
| - if b.initialSplits > 0 { | |
| - if forSize && !b.sorted { | |
| - if err := b.createInitialSplits(ctx); err != nil { | |
| - return err | |
| - } | |
| + if b.flushCounts.total == 0 && forSize && b.initialSplits != 0 && !b.sorted { | |
| + if err := b.createInitialSplits(ctx); err != nil { | |
| + return err | |
| } | |
| - // Disable doing initial splits going forward. | |
| - b.initialSplits = 0 | |
| } | |
| for i := range b.curBuf.entries { | |
| @@ -301,133 +275,51 @@ func (b *BufferingAdder) doFlush(ctx context.Context, forSize bool) error { | |
| dueToSize := b.sink.flushCounts.sstSize - before.sstSize | |
| log.Infof(ctx, | |
| - "%s adder flushing %s (%s buffered/%0.2gx) wrote %d SSTs (avg: %s) with %d for splits, %d for size, took %v", | |
| - b.name, | |
| - b.kvSize, | |
| - b.curBuf.MemSize, | |
| - float64(b.kvSize)/float64(b.curBuf.MemSize), | |
| - files, | |
| - sz(written/int64(files)), | |
| - dueToSplits, | |
| - dueToSize, | |
| - timing(timeutil.Since(beforeSort)), | |
| + "flushing %s buffer wrote %d SSTs (avg: %s) with %d for splits, %d for size, took %v", | |
| + sz(b.curBuf.MemSize), files, sz(written/int64(files)), dueToSplits, dueToSize, timeutil.Since(beforeSort), | |
| ) | |
| } | |
| - | |
| if log.V(4) { | |
| log.Infof(ctx, | |
| - "%s adder has ingested %s (%s); spent %s filling, %v sorting, %v flushing (%v sink, %v sending, %v splitting, %v scattering %v)", | |
| + "bulk adder %s has ingested %s, spent %v sorting and %v flushing (%v sending, %v splitting). Flushed %d times due to buffer (%s) size. Flushed chunked as %d files (%d after split-retries), %d due to ranges, %d due to sst size.", | |
| b.name, | |
| sz(b.sink.totalRows.DataSize), | |
| - sorted(b.sorted), | |
| - timing(b.flushCounts.totalFilling), | |
| - timing(b.flushCounts.totalSort), | |
| - timing(b.flushCounts.totalFlush), | |
| - timing(b.sink.flushCounts.flushWait), | |
| - timing(b.sink.flushCounts.sendWait), | |
| - timing(b.sink.flushCounts.splitWait), | |
| - timing(b.sink.flushCounts.scatterWait), | |
| - b.sink.flushCounts.scatterMoved, | |
| - ) | |
| - } | |
| - | |
| - if log.V(5) { | |
| - log.Infof(ctx, | |
| - "%s adder has flushed %d times due to buffer size (%s), chunked as %d files (%d for ranges, %d for sst size, +%d for split-retries)", | |
| - b.name, | |
| + b.flushCounts.totalSort, | |
| + b.flushCounts.totalFlush, | |
| + b.sink.flushCounts.sendWait, | |
| + b.sink.flushCounts.splitWait, | |
| b.flushCounts.bufferSize, | |
| sz(b.memAcc.Used()), | |
| - b.sink.flushCounts.total, | |
| - b.sink.flushCounts.split, | |
| - b.sink.flushCounts.sstSize, | |
| - b.sink.flushCounts.files-b.sink.flushCounts.total, | |
| + b.sink.flushCounts.total, b.sink.flushCounts.files, | |
| + b.sink.flushCounts.split, b.sink.flushCounts.sstSize, | |
| ) | |
| } | |
| - | |
| if b.onFlush != nil { | |
| b.onFlush(b.sink.GetBatchSummary()) | |
| } | |
| b.curBuf.Reset() | |
| - b.kvSize = 0 | |
| - b.lastFlush = timeutil.Now() | |
| return nil | |
| } | |
| func (b *BufferingAdder) createInitialSplits(ctx context.Context) error { | |
| - log.Infof(ctx, "%s adder creating up to %d initial splits from %d KVs in %s buffer", | |
| - b.name, b.initialSplits, b.curBuf.Len(), b.curBuf.MemSize) | |
| + targetSize := b.curBuf.Len() / b.initialSplits | |
| + log.Infof(ctx, "creating up to %d initial splits from %d keys in %s buffer", b.initialSplits, b.curBuf.Len(), sz(b.curBuf.MemSize)) | |
| hour := hlc.Timestamp{WallTime: timeutil.Now().Add(time.Hour).UnixNano()} | |
| - before := timeutil.Now() | |
| - | |
| - created := 0 | |
| - width := len(b.curBuf.entries) / b.initialSplits | |
| - for i := 0; i < b.initialSplits; i++ { | |
| - expire := hour | |
| - if i == 0 { | |
| - // If we over-split because our input is loosely ordered and we're just | |
| - // seeing a sample of the first span here vs a sample of all of it, then | |
| - // we may not fill enough for these splits to remain on their own. In that | |
| - // case we'd really prefer the other splits be merged away first rather | |
| - // than the first split, as it serves the important purpose of segregating | |
| - // this processor's span from the one below it when is being constantly | |
| - // re-scattered by that processor, so give the first split an extra hour. | |
| - expire = hour.Add(time.Hour.Nanoseconds(), 0) | |
| - } | |
| - splitAt := i * width | |
| - if splitAt >= len(b.curBuf.entries) { | |
| - break | |
| - } | |
| - // Typically we split at splitAt if, and only if, its range still includes | |
| - // the prior split, indicating no other processor is also splitting this | |
| - // span. However, for the first split, there is no prior split, so we can | |
| - // use the next split instead, as it too proves the range that need to be | |
| - // split still has enough "width" (i.e. between two splits) to indicate that | |
| - // another processor hasn't already split it. | |
| - predicateAt := splitAt - width | |
| - if predicateAt < 0 { | |
| - next := splitAt + width | |
| - if next > len(b.curBuf.entries)-1 { | |
| - next = len(b.curBuf.entries) - 1 | |
| - } | |
| - predicateAt = next | |
| - } | |
| - splitKey, err := keys.EnsureSafeSplitKey(b.curBuf.Key(splitAt)) | |
| - if err != nil { | |
| - log.Warningf(ctx, "failed to generate pre-split key for key %s", b.curBuf.Key(splitAt)) | |
| - continue | |
| - } | |
| - predicateKey := b.curBuf.Key(predicateAt) | |
| - log.VEventf(ctx, 1, "pre-splitting span %d of %d at %s", i, b.initialSplits, splitKey) | |
| - resp, err := b.sink.db.SplitAndScatter(ctx, splitKey, expire, predicateKey) | |
| - if err != nil { | |
| + for i := targetSize; i < b.curBuf.Len(); i += targetSize { | |
| + k := b.curBuf.Key(i) | |
| + prev := b.curBuf.Key(i - targetSize) | |
| + log.VEventf(ctx, 1, "splitting at key %d / %d: %s", i, b.curBuf.Len(), k) | |
| + if err := b.sink.db.SplitAndScatter(ctx, k, hour, prev); err != nil { | |
| // TODO(dt): a typed error would be nice here. | |
| if strings.Contains(err.Error(), "predicate") { | |
| - log.VEventf(ctx, 1, "%s adder split at %s rejected, had previously split and no longer included %s", | |
| - b.name, splitKey, predicateKey) | |
| + log.VEventf(ctx, 1, "split at %s rejected, had previously split and no longer included %s", k, prev) | |
| continue | |
| } | |
| return err | |
| } | |
| - | |
| - b.sink.flushCounts.splitWait += resp.Timing.Split | |
| - b.sink.flushCounts.scatterWait += resp.Timing.Scatter | |
| - if resp.ScatteredStats != nil { | |
| - moved := sz(resp.ScatteredStats.Total()) | |
| - b.sink.flushCounts.scatterMoved += moved | |
| - if resp.ScatteredStats.Total() > 0 { | |
| - log.VEventf(ctx, 1, "pre-split scattered %s in non-empty range %s", | |
| - moved, resp.ScatteredSpan) | |
| - } | |
| - } | |
| - created++ | |
| } | |
| - | |
| - log.Infof(ctx, "%s adder created %d initial splits in %v from %d keys in %s buffer", | |
| - b.name, created, timing(timeutil.Since(before)), b.curBuf.Len(), b.curBuf.MemSize) | |
| - | |
| - b.sink.initialSplitDone = true | |
| return nil | |
| } | |
| diff --git a/pkg/kv/bulk/kv_buf.go b/pkg/kv/bulk/kv_buf.go | |
| index 6ae5e91c7e..4e8f71f23e 100644 | |
| --- a/pkg/kv/bulk/kv_buf.go | |
| +++ b/pkg/kv/bulk/kv_buf.go | |
| @@ -26,7 +26,7 @@ import ( | |
| type kvBuf struct { | |
| entries []kvBufEntry | |
| slab []byte | |
| - MemSize sz // size of buffered data including per-entry overhead | |
| + MemSize int // size of buffered data including per-entry overhead | |
| } | |
| // each entry in the buffer has a key and value -- the actual bytes of these are | |
| @@ -57,7 +57,7 @@ func (b *kvBuf) append(k, v []byte) error { | |
| return errors.Errorf("length %d exceeds limit %d", len(v), maxLen) | |
| } | |
| - b.MemSize += sz(len(k) + len(v) + entryOverhead) | |
| + b.MemSize += len(k) + len(v) + entryOverhead | |
| var e kvBufEntry | |
| e.keySpan = uint64(len(b.slab)<<lenBits) | uint64(len(k)&lenMask) | |
| b.slab = append(b.slab, k...) | |
| diff --git a/pkg/kv/bulk/kv_buf_test.go b/pkg/kv/bulk/kv_buf_test.go | |
| index ae77662671..7e29efcc0e 100644 | |
| --- a/pkg/kv/bulk/kv_buf_test.go | |
| +++ b/pkg/kv/bulk/kv_buf_test.go | |
| @@ -26,7 +26,7 @@ type kvPair struct { | |
| value []byte | |
| } | |
| -func makeTestData(num int) (kvs []kvPair, totalSize sz) { | |
| +func makeTestData(num int) (kvs []kvPair, totalSize int) { | |
| kvs = make([]kvPair, num) | |
| r, _ := randutil.NewTestRand() | |
| alloc := make([]byte, num*500) | |
| @@ -41,7 +41,7 @@ func makeTestData(num int) (kvs []kvPair, totalSize sz) { | |
| alloc = alloc[len(kvs[i].key):] | |
| kvs[i].value = alloc[:randutil.RandIntInRange(r, 0, 1000)] | |
| alloc = alloc[len(kvs[i].value):] | |
| - totalSize += sz(len(kvs[i].key) + len(kvs[i].value)) | |
| + totalSize += len(kvs[i].key) + len(kvs[i].value) | |
| } | |
| return kvs, totalSize | |
| } | |
| @@ -63,7 +63,7 @@ func TestKvBuf(t *testing.T) { | |
| if expected, actual := len(src), b.Len(); expected != actual { | |
| t.Fatalf("expected len %d got %d", expected, actual) | |
| } | |
| - if expected, actual := totalSize+sz(len(src)*16), b.MemSize; expected != actual { | |
| + if expected, actual := totalSize+len(src)*16, b.MemSize; expected != actual { | |
| t.Fatalf("expected len %d got %d", expected, actual) | |
| } | |
| diff --git a/pkg/kv/bulk/setting.go b/pkg/kv/bulk/setting.go | |
| index c332ccf290..bb419dc94d 100644 | |
| --- a/pkg/kv/bulk/setting.go | |
| +++ b/pkg/kv/bulk/setting.go | |
| @@ -16,19 +16,21 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| ) | |
| -var ( | |
| - // IngestBatchSize controls the size of ingest ssts. | |
| - IngestBatchSize = settings.RegisterByteSizeSetting( | |
| +// IngestBatchSize is a cluster setting that controls the maximum size of the | |
| +// payload in an AddSSTable request. | |
| +var IngestBatchSize = func() *settings.ByteSizeSetting { | |
| + s := settings.RegisterByteSizeSetting( | |
| settings.TenantWritable, | |
| "kv.bulk_ingest.batch_size", | |
| "the maximum size of the payload in an AddSSTable request", | |
| 16<<20, | |
| ) | |
| -) | |
| + return s | |
| +}() | |
| -// ingestFileSize determines the target size files sent via AddSSTable requests. | |
| +// IngestFileSize determines the target size files sent via AddSSTable requests. | |
| // It returns the smaller of the IngestBatchSize and Raft command size settings. | |
| -func ingestFileSize(st *cluster.Settings) int64 { | |
| +func IngestFileSize(st *cluster.Settings) int64 { | |
| desiredSize := IngestBatchSize.Get(&st.SV) | |
| maxCommandSize := kvserver.MaxCommandSize.Get(&st.SV) | |
| if desiredSize > maxCommandSize { | |
| diff --git a/pkg/kv/bulk/sst_batcher.go b/pkg/kv/bulk/sst_batcher.go | |
| index d2663ae6eb..21e5302e3e 100644 | |
| --- a/pkg/kv/bulk/sst_batcher.go | |
| +++ b/pkg/kv/bulk/sst_batcher.go | |
| @@ -16,7 +16,6 @@ import ( | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/keys" | |
| - "github.com/cockroachdb/cockroach/pkg/kv" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvclient/rangecache" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverbase" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -29,6 +28,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| + "github.com/cockroachdb/redact" | |
| ) | |
| var ( | |
| @@ -39,13 +39,6 @@ var ( | |
| 400*1<<10, // 400 Kib | |
| ) | |
| - splitAfter = settings.RegisterByteSizeSetting( | |
| - settings.TenantWritable, | |
| - "bulkio.ingest.scatter_after_size", | |
| - "amount of data added to any one range after which a new range should be split off and scattered", | |
| - 48<<20, | |
| - ) | |
| - | |
| ingestDelay = settings.RegisterDurationSetting( | |
| settings.TenantWritable, | |
| "bulkio.ingest.flush_delay", | |
| @@ -57,23 +50,14 @@ var ( | |
| type sz int64 | |
| -func (b sz) String() string { return string(humanizeutil.IBytes(int64(b))) } | |
| -func (b sz) SafeValue() {} | |
| - | |
| -type timing time.Duration | |
| - | |
| -func (t timing) String() string { return time.Duration(t).Round(time.Second).String() } | |
| -func (t timing) SafeValue() {} | |
| - | |
| -type sorted bool | |
| +func (b sz) String() string { | |
| + return redact.StringWithoutMarkers(b) | |
| +} | |
| -func (t sorted) String() string { | |
| - if t { | |
| - return "sorted" | |
| - } | |
| - return "unsorted" | |
| +// SafeFormat implements the redact.SafeFormatter interface. | |
| +func (b sz) SafeFormat(w redact.SafePrinter, _ rune) { | |
| + w.Print(humanizeutil.IBytes(int64(b))) | |
| } | |
| -func (t sorted) SafeValue() {} | |
| // SSTBatcher is a helper for bulk-adding many KVs in chunks via AddSSTable. An | |
| // SSTBatcher can be handed KVs repeatedly and will make them into SSTs that are | |
| @@ -82,9 +66,11 @@ func (t sorted) SafeValue() {} | |
| // it to attempt to flush SSTs before they cross range boundaries to minimize | |
| // expensive on-split retries. | |
| type SSTBatcher struct { | |
| - db *kv.DB | |
| - rc *rangecache.RangeCache | |
| - settings *cluster.Settings | |
| + db SSTSender | |
| + rc *rangecache.RangeCache | |
| + settings *cluster.Settings | |
| + maxSize func() int64 | |
| + splitAfter func() int64 | |
| // disallowShadowingBelow is described on roachpb.AddSSTableRequest. | |
| disallowShadowingBelow hlc.Timestamp | |
| @@ -113,8 +99,6 @@ type SSTBatcher struct { | |
| // writeAtBatchTS is passed to the writeAtBatchTs argument to db.AddSStable. | |
| writeAtBatchTS bool | |
| - initialSplitDone bool | |
| - | |
| // The rest of the fields accumulated state as opposed to configuration. Some, | |
| // like totalRows, are accumulated _across_ batches and are not reset between | |
| // batches when Reset() is called. | |
| @@ -125,12 +109,8 @@ type SSTBatcher struct { | |
| sstSize int | |
| files int // a single flush might create multiple files. | |
| - scatterMoved sz | |
| - | |
| - flushWait time.Duration | |
| - sendWait time.Duration | |
| - splitWait time.Duration | |
| - scatterWait time.Duration | |
| + sendWait time.Duration | |
| + splitWait time.Duration | |
| } | |
| // Tracking for if we have "filled" a range in case we want to split/scatter. | |
| flushedToCurrentRange int64 | |
| @@ -155,14 +135,16 @@ type SSTBatcher struct { | |
| // MakeSSTBatcher makes a ready-to-use SSTBatcher. | |
| func MakeSSTBatcher( | |
| ctx context.Context, | |
| - db *kv.DB, | |
| + db SSTSender, | |
| settings *cluster.Settings, | |
| + flushBytes func() int64, | |
| disallowShadowingBelow hlc.Timestamp, | |
| writeAtBatchTs bool, | |
| ) (*SSTBatcher, error) { | |
| b := &SSTBatcher{ | |
| db: db, | |
| settings: settings, | |
| + maxSize: flushBytes, | |
| disallowShadowingBelow: disallowShadowingBelow, | |
| writeAtBatchTS: writeAtBatchTs, | |
| } | |
| @@ -173,9 +155,9 @@ func MakeSSTBatcher( | |
| // MakeStreamSSTBatcher creates a batcher configured to ingest duplicate keys | |
| // that might be received from a cluster to cluster stream. | |
| func MakeStreamSSTBatcher( | |
| - ctx context.Context, db *kv.DB, settings *cluster.Settings, | |
| + ctx context.Context, db SSTSender, settings *cluster.Settings, flushBytes func() int64, | |
| ) (*SSTBatcher, error) { | |
| - b := &SSTBatcher{db: db, settings: settings, ingestAll: true} | |
| + b := &SSTBatcher{db: db, settings: settings, maxSize: flushBytes, ingestAll: true} | |
| err := b.Reset(ctx) | |
| return b, err | |
| } | |
| @@ -304,7 +286,7 @@ func (b *SSTBatcher) flushIfNeeded(ctx context.Context, nextKey roachpb.Key) err | |
| return b.Reset(ctx) | |
| } | |
| - if b.sstWriter.DataSize >= ingestFileSize(b.settings) { | |
| + if b.sstWriter.DataSize >= b.maxSize() { | |
| if err := b.doFlush(ctx, sizeFlush, nextKey); err != nil { | |
| return err | |
| } | |
| @@ -329,8 +311,6 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| if b.sstWriter.DataSize == 0 { | |
| return nil | |
| } | |
| - beforeFlush := timeutil.Now() | |
| - | |
| b.flushCounts.total++ | |
| if delay := ingestDelay.Get(&b.settings.SV); delay != 0 { | |
| @@ -344,7 +324,7 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| } | |
| } | |
| - hour := hlc.Timestamp{WallTime: beforeFlush.Add(time.Hour).UnixNano()} | |
| + hour := hlc.Timestamp{WallTime: timeutil.Now().Add(time.Hour).UnixNano()} | |
| start := roachpb.Key(append([]byte(nil), b.batchStartKey...)) | |
| // The end key of the WriteBatch request is exclusive, but batchEndKey is | |
| @@ -353,7 +333,7 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| size := b.sstWriter.DataSize | |
| if reason == sizeFlush { | |
| - log.VEventf(ctx, 3, "flushing %s SST due to size > %s", sz(size), sz(ingestFileSize(b.settings))) | |
| + log.VEventf(ctx, 3, "flushing %s SST due to size > %s", sz(size), sz(b.maxSize())) | |
| b.flushCounts.sstSize++ | |
| // On first flush, if it is due to size, we introduce one split at the start | |
| @@ -363,30 +343,17 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| // minimizes impact on other adders (e.g. causing extra SST splitting). | |
| // | |
| // We only do this splitting if the caller expects the sst_batcher to | |
| - // split and scatter the data as it ingests it i.e. splitAfter > 0. | |
| - if !b.initialSplitDone && b.flushCounts.total == 1 && splitAfter.Get(&b.settings.SV) > 0 { | |
| + // split and scatter the data as it ingests it (which is the case when | |
| + // splitAfter) is set. | |
| + if b.flushCounts.total == 1 && b.splitAfter != nil { | |
| if splitAt, err := keys.EnsureSafeSplitKey(start); err != nil { | |
| - log.Warningf(ctx, "failed to generate split key to separate ingestion span: %v", err) | |
| + log.Warningf(ctx, "%v", err) | |
| } else { | |
| - if log.V(1) { | |
| - log.Infof(ctx, "splitting on first flush to separate ingestion span using key %v", start) | |
| - } | |
| // NB: Passing 'hour' here is technically illegal until 19.2 is | |
| // active, but the value will be ignored before that, and we don't | |
| // have access to the cluster version here. | |
| - reply, err := b.db.SplitAndScatter(ctx, splitAt, hour) | |
| - if err != nil { | |
| - log.Warningf(ctx, "failed ot split at first key to separate ingestion span: %v", err) | |
| - } else { | |
| - b.flushCounts.splitWait += reply.Timing.Split | |
| - b.flushCounts.scatterWait += reply.Timing.Scatter | |
| - if reply.ScatteredStats != nil { | |
| - moved := sz(reply.ScatteredStats.Total()) | |
| - b.flushCounts.scatterMoved += moved | |
| - if moved > 0 { | |
| - log.VEventf(ctx, 1, "starting split scattered %s in non-empty range %s", moved, reply.ScatteredSpan) | |
| - } | |
| - } | |
| + if err := b.db.SplitAndScatter(ctx, splitAt, hour); err != nil { | |
| + log.Warningf(ctx, "%v", err) | |
| } | |
| } | |
| } | |
| @@ -426,28 +393,21 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| b.lastFlushKey = append(b.lastFlushKey[:0], b.flushKey...) | |
| b.flushedToCurrentRange = size | |
| } | |
| - if splitSize := splitAfter.Get(&b.settings.SV); splitSize > 0 { | |
| - if b.flushedToCurrentRange > splitSize && nextKey != nil { | |
| + if b.splitAfter != nil { | |
| + if splitAfter := b.splitAfter(); b.flushedToCurrentRange > splitAfter && nextKey != nil { | |
| if splitAt, err := keys.EnsureSafeSplitKey(nextKey); err != nil { | |
| log.Warningf(ctx, "%v", err) | |
| } else { | |
| + beforeSplit := timeutil.Now() | |
| + | |
| log.VEventf(ctx, 2, "%s added since last split, splitting/scattering for next range at %v", sz(b.flushedToCurrentRange), end) | |
| - reply, err := b.db.SplitAndScatter(ctx, splitAt, hour) | |
| - if err != nil { | |
| + // NB: Passing 'hour' here is technically illegal until 19.2 is | |
| + // active, but the value will be ignored before that, and we don't | |
| + // have access to the cluster version here. | |
| + if err := b.db.SplitAndScatter(ctx, splitAt, hour); err != nil { | |
| log.Warningf(ctx, "failed to split and scatter during ingest: %+v", err) | |
| } | |
| - b.flushCounts.splitWait += reply.Timing.Split | |
| - b.flushCounts.scatterWait += reply.Timing.Scatter | |
| - if reply.ScatteredStats != nil { | |
| - moved := sz(reply.ScatteredStats.Total()) | |
| - b.flushCounts.scatterMoved += moved | |
| - if moved > 0 { | |
| - // This is unexpected, since 'filling' a range without hitting a | |
| - // an existing split suggests non-overlapping input, so we expect | |
| - // our still-to-fill RHS to be empty and cheap to move. | |
| - log.VEventf(ctx, 1, "filled-range split scattered %s in non-empty range %s", moved, reply.ScatteredSpan) | |
| - } | |
| - } | |
| + b.flushCounts.splitWait += timeutil.Since(beforeSplit) | |
| } | |
| b.flushedToCurrentRange = 0 | |
| } | |
| @@ -456,7 +416,6 @@ func (b *SSTBatcher) doFlush(ctx context.Context, reason int, nextKey roachpb.Ke | |
| b.rowCounter.DataSize += b.sstWriter.DataSize | |
| b.totalRows.Add(b.rowCounter.BulkOpSummary) | |
| - b.flushCounts.flushWait += timeutil.Since(beforeFlush) | |
| return nil | |
| } | |
| @@ -475,6 +434,39 @@ func (b *SSTBatcher) GetSummary() roachpb.BulkOpSummary { | |
| return b.totalRows | |
| } | |
| +// SSTSender is an interface to send SST data to an engine. | |
| +type SSTSender interface { | |
| + AddSSTable( | |
| + ctx context.Context, | |
| + begin, end interface{}, | |
| + data []byte, | |
| + disallowConflicts bool, | |
| + disallowShadowing bool, | |
| + disallowShadowingBelow hlc.Timestamp, | |
| + stats *enginepb.MVCCStats, | |
| + ingestAsWrites bool, | |
| + batchTs hlc.Timestamp, | |
| + ) error | |
| + | |
| + AddSSTableAtBatchTimestamp( | |
| + ctx context.Context, | |
| + begin, end interface{}, | |
| + data []byte, | |
| + disallowConflicts bool, | |
| + disallowShadowing bool, | |
| + disallowShadowingBelow hlc.Timestamp, | |
| + stats *enginepb.MVCCStats, | |
| + ingestAsWrites bool, | |
| + batchTs hlc.Timestamp, | |
| + ) (hlc.Timestamp, error) | |
| + | |
| + SplitAndScatter( | |
| + ctx context.Context, key roachpb.Key, expirationTime hlc.Timestamp, predicateKeys ...roachpb.Key, | |
| + ) error | |
| + | |
| + Clock() *hlc.Clock | |
| +} | |
| + | |
| type sstSpan struct { | |
| start, end roachpb.Key | |
| sstBytes []byte | |
| @@ -487,7 +479,7 @@ type sstSpan struct { | |
| // for each side of the split in the error, and each are retried. | |
| func AddSSTable( | |
| ctx context.Context, | |
| - db *kv.DB, | |
| + db SSTSender, | |
| start, end roachpb.Key, | |
| sstBytes []byte, | |
| disallowShadowingBelow hlc.Timestamp, | |
| @@ -561,11 +553,6 @@ func AddSSTable( | |
| log.VEventf(ctx, 3, "adding %s AddSSTable [%s,%s) took %v", sz(len(item.sstBytes)), item.start, item.end, timeutil.Since(before)) | |
| return nil | |
| } | |
| - // Retry on AmbiguousResult. | |
| - if errors.HasType(err, (*roachpb.AmbiguousResultError)(nil)) { | |
| - log.Warningf(ctx, "addsstable [%s,%s) attempt %d failed: %+v", start, end, i, err) | |
| - continue | |
| - } | |
| // This range has split -- we need to split the SST to try again. | |
| if m := (*roachpb.RangeKeyMismatchError)(nil); errors.As(err, &m) { | |
| // TODO(andrei): We just use the first of m.Ranges; presumably we | |
| @@ -594,6 +581,11 @@ func AddSSTable( | |
| work = append([]*sstSpan{left, right}, work...) | |
| return nil | |
| } | |
| + // Retry on AmbiguousResult. | |
| + if errors.HasType(err, (*roachpb.AmbiguousResultError)(nil)) { | |
| + log.Warningf(ctx, "addsstable [%s,%s) attempt %d failed: %+v", start, end, i, err) | |
| + continue | |
| + } | |
| } | |
| return errors.Wrapf(err, "addsstable [%s,%s)", item.start, item.end) | |
| }(); err != nil { | |
| @@ -612,7 +604,7 @@ func AddSSTable( | |
| // passed in is over the top level SST passed into AddSSTTable(). | |
| func createSplitSSTable( | |
| ctx context.Context, | |
| - db *kv.DB, | |
| + db SSTSender, | |
| start, splitKey roachpb.Key, | |
| disallowShadowingBelow hlc.Timestamp, | |
| iter storage.SimpleMVCCIterator, | |
| diff --git a/pkg/kv/bulk/sst_batcher_test.go b/pkg/kv/bulk/sst_batcher_test.go | |
| index ac1482f288..ae0bd0f99c 100644 | |
| --- a/pkg/kv/bulk/sst_batcher_test.go | |
| +++ b/pkg/kv/bulk/sst_batcher_test.go | |
| @@ -132,7 +132,7 @@ func runTestImport(t *testing.T, batchSizeValue int64) { | |
| ts := hlc.Timestamp{WallTime: 100} | |
| b, err := bulk.MakeBulkAdder( | |
| - ctx, kvDB, mockCache, s.ClusterSettings(), ts, kvserverbase.BulkAdderOptions{MinBufferSize: batchSize()}, nil, /* bulkMon */ | |
| + ctx, kvDB, mockCache, s.ClusterSettings(), ts, kvserverbase.BulkAdderOptions{MinBufferSize: batchSize(), SSTSize: batchSize}, nil, /* bulkMon */ | |
| ) | |
| if err != nil { | |
| t.Fatal(err) | |
| diff --git a/pkg/kv/db.go b/pkg/kv/db.go | |
| index 231252f00b..05e834a1a2 100644 | |
| --- a/pkg/kv/db.go | |
| +++ b/pkg/kv/db.go | |
| @@ -14,11 +14,9 @@ import ( | |
| "context" | |
| "fmt" | |
| "strings" | |
| - "time" | |
| "github.com/cockroachdb/cockroach/pkg/base" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/sql/sessiondatapb" | |
| "github.com/cockroachdb/cockroach/pkg/storage/enginepb" | |
| "github.com/cockroachdb/cockroach/pkg/util/admission" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| @@ -26,7 +24,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/protoutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/retry" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| - "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| ) | |
| @@ -585,55 +582,24 @@ func (db *DB) AdminSplit( | |
| return getOneErr(db.Run(ctx, b), b) | |
| } | |
| -// SplitAndScatterResult carries wraps information about the SplitAndScatter | |
| -// call, including how long each step took or stats for range scattered. | |
| -type SplitAndScatterResult struct { | |
| - // Timing indicates how long each step in this multi-step call took. | |
| - Timing struct { | |
| - Split time.Duration | |
| - Scatter time.Duration | |
| - } | |
| - // Stats describe the scattered range, as returned by the AdminScatter call. | |
| - ScatteredStats *enginepb.MVCCStats | |
| - ScatteredSpan roachpb.Span | |
| -} | |
| - | |
| // SplitAndScatter is a helper that wraps AdminSplit + AdminScatter. | |
| func (db *DB) SplitAndScatter( | |
| ctx context.Context, key roachpb.Key, expirationTime hlc.Timestamp, predicateKeys ...roachpb.Key, | |
| -) (SplitAndScatterResult, error) { | |
| - beforeSplit := timeutil.Now() | |
| +) error { | |
| b := &Batch{} | |
| b.adminSplit(key, expirationTime, predicateKeys) | |
| if err := getOneErr(db.Run(ctx, b), b); err != nil { | |
| - return SplitAndScatterResult{}, err | |
| + return err | |
| } | |
| - beforeScatter := timeutil.Now() | |
| scatterReq := &roachpb.AdminScatterRequest{ | |
| RequestHeader: roachpb.RequestHeaderFromSpan(roachpb.Span{Key: key, EndKey: key.Next()}), | |
| RandomizeLeases: true, | |
| } | |
| - raw, pErr := SendWrapped(ctx, db.NonTransactionalSender(), scatterReq) | |
| - if pErr != nil { | |
| - return SplitAndScatterResult{}, pErr.GoError() | |
| - } | |
| - reply := SplitAndScatterResult{} | |
| - reply.Timing.Split = beforeScatter.Sub(beforeSplit) | |
| - reply.Timing.Scatter = timeutil.Since(beforeScatter) | |
| - resp, ok := raw.(*roachpb.AdminScatterResponse) | |
| - if !ok { | |
| - return reply, errors.Errorf("unexpected response of type %T for AdminScatter", raw) | |
| - } | |
| - reply.ScatteredStats = resp.MVCCStats | |
| - if len(resp.RangeInfos) > 0 { | |
| - reply.ScatteredSpan = roachpb.Span{ | |
| - Key: resp.RangeInfos[0].Desc.StartKey.AsRawKey(), | |
| - EndKey: resp.RangeInfos[0].Desc.EndKey.AsRawKey(), | |
| - } | |
| + if _, pErr := SendWrapped(ctx, db.NonTransactionalSender(), scatterReq); pErr != nil { | |
| + return pErr.GoError() | |
| } | |
| - | |
| - return reply, nil | |
| + return nil | |
| } | |
| // AdminUnsplit removes the sticky bit of the range specified by splitKey. | |
| @@ -910,25 +876,6 @@ func (db *DB) Txn(ctx context.Context, retryable func(context.Context, *Txn) err | |
| return runTxn(ctx, txn, retryable) | |
| } | |
| -// TxnWithSteppingEnabled is the same Txn, but represents a request originating | |
| -// from SQL and has stepping enabled and quality of service set. | |
| -func (db *DB) TxnWithSteppingEnabled( | |
| - ctx context.Context, | |
| - qualityOfService sessiondatapb.QoSLevel, | |
| - retryable func(context.Context, *Txn) error, | |
| -) error { | |
| - // TODO(radu): we should open a tracing Span here (we need to figure out how | |
| - // to use the correct tracer). | |
| - | |
| - // Observed timestamps don't work with multi-tenancy. See: | |
| - // | |
| - // https://github.com/cockroachdb/cockroach/issues/48008 | |
| - nodeID, _ := db.ctx.NodeID.OptionalNodeID() // zero if not available | |
| - txn := NewTxnWithSteppingEnabled(ctx, db, nodeID, qualityOfService) | |
| - txn.SetDebugName("unnamed") | |
| - return runTxn(ctx, txn, retryable) | |
| -} | |
| - | |
| // TxnRootKV is the same as Txn, but specifically represents a request | |
| // originating within KV, and that is at the root of the tree of requests. For | |
| // KV usage that should be subject to admission control. Do not use this for | |
| diff --git a/pkg/kv/kvclient/kvcoord/BUILD.bazel b/pkg/kv/kvclient/kvcoord/BUILD.bazel | |
| index 817e23cca3..3f22e3713e 100644 | |
| --- a/pkg/kv/kvclient/kvcoord/BUILD.bazel | |
| +++ b/pkg/kv/kvclient/kvcoord/BUILD.bazel | |
| @@ -134,7 +134,6 @@ go_test( | |
| ], | |
| data = glob(["testdata/**"]), | |
| embed = [":kvcoord"], | |
| - shard_count = 16, | |
| tags = ["no-remote"], | |
| deps = [ | |
| "//build/bazelutil:noop", | |
| diff --git a/pkg/kv/kvclient/kvcoord/dist_sender.go b/pkg/kv/kvclient/kvcoord/dist_sender.go | |
| index 3b7229ac65..8d3bee67d1 100644 | |
| --- a/pkg/kv/kvclient/kvcoord/dist_sender.go | |
| +++ b/pkg/kv/kvclient/kvcoord/dist_sender.go | |
| @@ -609,7 +609,7 @@ func (ds *DistSender) getRoutingInfo( | |
| } | |
| if !containsFn(returnToken.Desc(), descKey) { | |
| log.Fatalf(ctx, "programming error: range resolution returning non-matching descriptor: "+ | |
| - "desc: %s, key: %s, reverse: %t", returnToken.Desc(), descKey, redact.Safe(useReverseScan)) | |
| + "desc: %s, key: %s, reverse: %t", returnToken.Desc(), descKey, log.Safe(useReverseScan)) | |
| } | |
| } | |
| @@ -764,7 +764,7 @@ func (ds *DistSender) Send( | |
| // We already verified above that the batch contains only scan requests of the same type. | |
| // Such a batch should never need splitting. | |
| log.Fatalf(ctx, "batch with MaxSpanRequestKeys=%d, TargetBytes=%d needs splitting", | |
| - redact.Safe(ba.MaxSpanRequestKeys), redact.Safe(ba.TargetBytes)) | |
| + log.Safe(ba.MaxSpanRequestKeys), log.Safe(ba.TargetBytes)) | |
| } | |
| var singleRplChunk [1]*roachpb.BatchResponse | |
| rplChunks := singleRplChunk[:0:1] | |
| @@ -2212,18 +2212,16 @@ func (ds *DistSender) sendToReplicas( | |
| // Has the caller given up? | |
| if ctx.Err() != nil { | |
| + reportedErr := errors.Wrap(ctx.Err(), "context done during DistSender.Send") | |
| + log.Eventf(ctx, "%v", reportedErr) | |
| + if ambiguousError != nil { | |
| + return nil, roachpb.NewAmbiguousResultErrorf(reportedErr.Error()) | |
| + } | |
| // Don't consider this a sendError, because sendErrors indicate that we | |
| // were unable to reach a replica that could serve the request, and they | |
| // cause range cache evictions. Context cancellations just mean the | |
| // sender changed its mind or the request timed out. | |
| - | |
| - if ambiguousError != nil { | |
| - err = roachpb.NewAmbiguousResultError(errors.Wrapf(ambiguousError, "context done during DistSender.Send")) | |
| - } else { | |
| - err = errors.Wrap(ctx.Err(), "aborted during DistSender.Send") | |
| - } | |
| - log.Eventf(ctx, "%v", err) | |
| - return nil, err | |
| + return nil, errors.Wrap(ctx.Err(), "aborted during DistSender.Send") | |
| } | |
| } | |
| } | |
| diff --git a/pkg/kv/kvclient/kvcoord/dist_sender_test.go b/pkg/kv/kvclient/kvcoord/dist_sender_test.go | |
| index 779ead4e98..338077d063 100644 | |
| --- a/pkg/kv/kvclient/kvcoord/dist_sender_test.go | |
| +++ b/pkg/kv/kvclient/kvcoord/dist_sender_test.go | |
| @@ -3151,14 +3151,14 @@ func TestParallelCommitsDetectIntentMissingCause(t *testing.T) { | |
| queryTxnFn: func() (roachpb.TransactionStatus, bool, error) { | |
| return roachpb.ABORTED, txnRecordSynthesized, nil | |
| }, | |
| - expErr: "result is ambiguous: intent missing and record aborted", | |
| + expErr: "result is ambiguous (intent missing and record aborted)", | |
| }, | |
| { | |
| name: "QueryTxn error, unresolved ambiguity", | |
| queryTxnFn: func() (roachpb.TransactionStatus, bool, error) { | |
| return 0, false, errors.New("unable to query txn") | |
| }, | |
| - expErr: "result is ambiguous: error=unable to query txn [intent missing]", | |
| + expErr: "result is ambiguous (error=unable to query txn [intent missing])", | |
| }, | |
| } | |
| for _, test := range testCases { | |
| diff --git a/pkg/kv/kvclient/kvcoord/txn_coord_sender_test.go b/pkg/kv/kvclient/kvcoord/txn_coord_sender_test.go | |
| index aa178d762c..27ada82dfd 100644 | |
| --- a/pkg/kv/kvclient/kvcoord/txn_coord_sender_test.go | |
| +++ b/pkg/kv/kvclient/kvcoord/txn_coord_sender_test.go | |
| @@ -670,7 +670,7 @@ func TestTxnCoordSenderGCWithAmbiguousResultErr(t *testing.T) { | |
| testutils.RunTrueAndFalse(t, "errOnFirst", func(t *testing.T, errOnFirst bool) { | |
| key := roachpb.Key("a") | |
| - are := roachpb.NewAmbiguousResultErrorf("very ambiguous") | |
| + are := roachpb.NewAmbiguousResultError("very ambiguous") | |
| knobs := &kvserver.StoreTestingKnobs{ | |
| TestingResponseFilter: func(ctx context.Context, ba roachpb.BatchRequest, br *roachpb.BatchResponse) *roachpb.Error { | |
| for _, req := range ba.Requests { | |
| diff --git a/pkg/kv/kvclient/kvtenant/connector.go b/pkg/kv/kvclient/kvtenant/connector.go | |
| index 3db120cc51..71756dea4c 100644 | |
| --- a/pkg/kv/kvclient/kvtenant/connector.go | |
| +++ b/pkg/kv/kvclient/kvtenant/connector.go | |
| @@ -61,11 +61,6 @@ type Connector interface { | |
| // primitives. | |
| serverpb.RegionsServer | |
| - // TenantStatusServer is the subset of the serverpb.StatusInterface that is | |
| - // used by the SQL system to query for debug information, such as tenant-specific | |
| - // range reports. | |
| - serverpb.TenantStatusServer | |
| - | |
| // TokenBucketProvider provides access to the tenant cost control token | |
| // bucket. | |
| TokenBucketProvider | |
| diff --git a/pkg/kv/kvclient/rangefeed/BUILD.bazel b/pkg/kv/kvclient/rangefeed/BUILD.bazel | |
| index a419196ed2..55201509a2 100644 | |
| --- a/pkg/kv/kvclient/rangefeed/BUILD.bazel | |
| +++ b/pkg/kv/kvclient/rangefeed/BUILD.bazel | |
| @@ -32,7 +32,6 @@ go_library( | |
| "//pkg/util/timeutil", | |
| "@com_github_cockroachdb_errors//:errors", | |
| "@com_github_cockroachdb_logtags//:logtags", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| ], | |
| ) | |
| diff --git a/pkg/kv/kvclient/rangefeed/rangefeed.go b/pkg/kv/kvclient/rangefeed/rangefeed.go | |
| index 85e9e8a7e9..fb377e484e 100644 | |
| --- a/pkg/kv/kvclient/rangefeed/rangefeed.go | |
| +++ b/pkg/kv/kvclient/rangefeed/rangefeed.go | |
| @@ -31,7 +31,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| "github.com/cockroachdb/logtags" | |
| - "github.com/cockroachdb/redact" | |
| ) | |
| //go:generate mockgen -destination=mocks_generated_test.go --package=rangefeed . DB | |
| @@ -300,7 +299,7 @@ func (f *RangeFeed) run(ctx context.Context, frontier *span.Frontier) { | |
| } | |
| if err != nil && ctx.Err() == nil && restartLogEvery.ShouldLog() { | |
| log.Warningf(ctx, "rangefeed failed %d times, restarting: %v", | |
| - redact.Safe(i), err) | |
| + log.Safe(i), err) | |
| } | |
| if ctx.Err() != nil { | |
| log.VEventf(ctx, 1, "exiting rangefeed") | |
| diff --git a/pkg/kv/kvnemesis/validator_test.go b/pkg/kv/kvnemesis/validator_test.go | |
| index da549ba68b..5f8eb2db3f 100644 | |
| --- a/pkg/kv/kvnemesis/validator_test.go | |
| +++ b/pkg/kv/kvnemesis/validator_test.go | |
| @@ -163,32 +163,32 @@ func TestValidate(t *testing.T) { | |
| }, | |
| { | |
| name: "one ambiguous put with successful write", | |
| - steps: []Step{step(withResult(put(`a`, `v1`), roachpb.NewAmbiguousResultError(errors.New("boom"))))}, | |
| + steps: []Step{step(withResult(put(`a`, `v1`), roachpb.NewAmbiguousResultError(``)))}, | |
| kvs: kvs(kv(`a`, 1, `v1`)), | |
| expected: nil, | |
| }, | |
| { | |
| name: "one ambiguous delete with successful write", | |
| - steps: []Step{step(withResult(del(`a`), roachpb.NewAmbiguousResultError(errors.New("boom"))))}, | |
| + steps: []Step{step(withResult(del(`a`), roachpb.NewAmbiguousResultError(``)))}, | |
| kvs: kvs(tombstone(`a`, 1)), | |
| expected: []string{`unable to validate delete operations in ambiguous transactions: [d]"a":missing-><nil>`}, | |
| }, | |
| { | |
| name: "one ambiguous put with failed write", | |
| - steps: []Step{step(withResult(put(`a`, `v1`), roachpb.NewAmbiguousResultError(errors.New("boom"))))}, | |
| + steps: []Step{step(withResult(put(`a`, `v1`), roachpb.NewAmbiguousResultError(``)))}, | |
| kvs: nil, | |
| expected: nil, | |
| }, | |
| { | |
| name: "one ambiguous delete with failed write", | |
| - steps: []Step{step(withResult(del(`a`), roachpb.NewAmbiguousResultError(errors.New("boom"))))}, | |
| + steps: []Step{step(withResult(del(`a`), roachpb.NewAmbiguousResultError(``)))}, | |
| kvs: nil, | |
| expected: nil, | |
| }, | |
| { | |
| name: "one ambiguous delete with failed write before a later committed delete", | |
| steps: []Step{ | |
| - step(withResult(del(`a`), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + step(withResult(del(`a`), roachpb.NewAmbiguousResultError(``))), | |
| step(withResult(del(`a`), nil)), | |
| }, | |
| kvs: kvs(tombstone(`a`, 1)), | |
| @@ -511,7 +511,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(put(`b`, `v2`), nil), | |
| - ), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: kvs(kv(`a`, 1, `v1`), kv(`b`, 1, `v2`)), | |
| expected: nil, | |
| @@ -522,7 +522,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(del(`b`), nil), | |
| - ), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: kvs(kv(`a`, 1, `v1`), tombstone(`b`, 1)), | |
| // TODO(sarkesian): If able to determine the tombstone resulting from a | |
| @@ -538,7 +538,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(put(`b`, `v2`), nil), | |
| - ), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: nil, | |
| expected: nil, | |
| @@ -549,7 +549,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(del(`b`), nil), | |
| - ), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: nil, | |
| expected: nil, | |
| @@ -560,7 +560,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(put(`b`, `v2`), nil), | |
| - ), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: kvs(kv(`a`, 1, `v1`), kv(`b`, 2, `v2`)), | |
| expected: []string{ | |
| @@ -573,7 +573,7 @@ func TestValidate(t *testing.T) { | |
| step(withResult(withTimestamp(closureTxn(ClosureTxnType_Commit, | |
| withResult(put(`a`, `v1`), nil), | |
| withResult(del(`b`), nil), | |
| - ), 2), roachpb.NewAmbiguousResultError(errors.New("boom")))), | |
| + ), 2), roachpb.NewAmbiguousResultError(``))), | |
| }, | |
| kvs: kvs(kv(`a`, 1, `v1`), tombstone(`b`, 2)), | |
| // TODO(sarkesian): If able to determine the tombstone resulting from a | |
| diff --git a/pkg/kv/kvserver/BUILD.bazel b/pkg/kv/kvserver/BUILD.bazel | |
| index 8e8c388fb3..e24b2673dc 100644 | |
| --- a/pkg/kv/kvserver/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/BUILD.bazel | |
| @@ -37,6 +37,7 @@ go_library( | |
| "replica_backpressure.go", | |
| "replica_batch_updates.go", | |
| "replica_circuit_breaker.go", | |
| + "replica_circuit_breaker_cancelstorage.go", | |
| "replica_closedts.go", | |
| "replica_command.go", | |
| "replica_consistency.go", | |
| @@ -123,7 +124,6 @@ go_library( | |
| "//pkg/kv/kvserver/closedts/sidetransport", | |
| "//pkg/kv/kvserver/closedts/tracker", | |
| "//pkg/kv/kvserver/concurrency", | |
| - "//pkg/kv/kvserver/concurrency/poison", | |
| "//pkg/kv/kvserver/constraint", | |
| "//pkg/kv/kvserver/gc", | |
| "//pkg/kv/kvserver/idalloc", | |
| @@ -132,6 +132,8 @@ go_library( | |
| "//pkg/kv/kvserver/kvserverpb", | |
| "//pkg/kv/kvserver/liveness", | |
| "//pkg/kv/kvserver/liveness/livenesspb", | |
| + "//pkg/kv/kvserver/protectedts", | |
| + "//pkg/kv/kvserver/protectedts/ptpb", | |
| "//pkg/kv/kvserver/raftentry", | |
| "//pkg/kv/kvserver/rangefeed", | |
| "//pkg/kv/kvserver/rditer", | |
| @@ -225,6 +227,7 @@ go_test( | |
| "client_rangefeed_test.go", | |
| "client_relocate_range_test.go", | |
| "client_replica_backpressure_test.go", | |
| + "client_replica_circuit_breaker_bench_test.go", | |
| "client_replica_circuit_breaker_test.go", | |
| "client_replica_gc_test.go", | |
| "client_replica_test.go", | |
| @@ -304,8 +307,8 @@ go_test( | |
| ], | |
| data = glob(["testdata/**"]), | |
| embed = [":kvserver"], | |
| - shard_count = 8, | |
| - tags = ["cpu:2"], | |
| + shard_count = 4, | |
| + tags = ["cpu:4"], | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/cli/exit", | |
| @@ -333,8 +336,10 @@ go_test( | |
| "//pkg/kv/kvserver/kvserverpb", | |
| "//pkg/kv/kvserver/liveness", | |
| "//pkg/kv/kvserver/liveness/livenesspb", | |
| + "//pkg/kv/kvserver/protectedts", | |
| "//pkg/kv/kvserver/protectedts/ptpb", | |
| "//pkg/kv/kvserver/protectedts/ptstorage", | |
| + "//pkg/kv/kvserver/protectedts/ptverifier", | |
| "//pkg/kv/kvserver/raftentry", | |
| "//pkg/kv/kvserver/rditer", | |
| "//pkg/kv/kvserver/readsummary/rspb", | |
| @@ -355,7 +360,6 @@ go_test( | |
| "//pkg/server/telemetry", | |
| "//pkg/settings/cluster", | |
| "//pkg/spanconfig", | |
| - "//pkg/spanconfig/spanconfigptsreader", | |
| "//pkg/spanconfig/spanconfigstore", | |
| "//pkg/sql", | |
| "//pkg/sql/catalog/bootstrap", | |
| diff --git a/pkg/kv/kvserver/allocator.go b/pkg/kv/kvserver/allocator.go | |
| index e756660ed2..7eb8af2d03 100644 | |
| --- a/pkg/kv/kvserver/allocator.go | |
| +++ b/pkg/kv/kvserver/allocator.go | |
| @@ -185,17 +185,6 @@ const ( | |
| nonVoterTarget | |
| ) | |
| -// replicaStatus represents whether a replica is currently alive, | |
| -// dead or decommissioning. | |
| -type replicaStatus int | |
| - | |
| -const ( | |
| - _ replicaStatus = iota | |
| - alive | |
| - dead | |
| - decommissioning | |
| -) | |
| - | |
| // AddChangeType returns the roachpb.ReplicaChangeType corresponding to the | |
| // given targetReplicaType. | |
| // | |
| diff --git a/pkg/kv/kvserver/batcheval/BUILD.bazel b/pkg/kv/kvserver/batcheval/BUILD.bazel | |
| index cb85b4c9d2..bc5d66ddb7 100644 | |
| --- a/pkg/kv/kvserver/batcheval/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/batcheval/BUILD.bazel | |
| @@ -86,7 +86,6 @@ go_library( | |
| "//pkg/util/tracing", | |
| "//pkg/util/uuid", | |
| "@com_github_cockroachdb_errors//:errors", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| "@com_github_gogo_protobuf//types", | |
| "@com_github_kr_pretty//:pretty", | |
| "@org_golang_x_time//rate", | |
| diff --git a/pkg/kv/kvserver/batcheval/cmd_push_txn.go b/pkg/kv/kvserver/batcheval/cmd_push_txn.go | |
| index ab7c60ddf6..fd8e78eac7 100644 | |
| --- a/pkg/kv/kvserver/batcheval/cmd_push_txn.go | |
| +++ b/pkg/kv/kvserver/batcheval/cmd_push_txn.go | |
| @@ -24,7 +24,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| ) | |
| func init() { | |
| @@ -269,10 +268,10 @@ func PushTxn( | |
| s = "failed to push" | |
| } | |
| log.Infof(ctx, "%s %s (push type=%s) %s: %s (pushee last active: %s)", | |
| - args.PusherTxn.Short(), redact.Safe(s), | |
| - redact.Safe(pushType), | |
| + args.PusherTxn.Short(), log.Safe(s), | |
| + log.Safe(pushType), | |
| args.PusheeTxn.Short(), | |
| - redact.Safe(reason), | |
| + log.Safe(reason), | |
| reply.PusheeTxn.LastActive()) | |
| } | |
| diff --git a/pkg/kv/kvserver/client_merge_test.go b/pkg/kv/kvserver/client_merge_test.go | |
| index c2e662ceae..c42254439d 100644 | |
| --- a/pkg/kv/kvserver/client_merge_test.go | |
| +++ b/pkg/kv/kvserver/client_merge_test.go | |
| @@ -417,7 +417,6 @@ func TestStoreRangeMergeTimestampCache(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| skip.UnderShort(t) | |
| - skip.UnderRace(t) | |
| testutils.RunTrueAndFalse(t, "disjoint-leaseholders", func(t *testing.T, disjointLeaseholders bool) { | |
| testutils.RunTrueAndFalse(t, "through-snapshot", func(t *testing.T, throughSnapshot bool) { | |
| @@ -4466,8 +4465,6 @@ func TestMergeQueueSeesNonVoters(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| - skip.UnderRace(t) | |
| - | |
| type test struct { | |
| name string | |
| leftVoters, rightVoters, leftNonVoters, rightNonVoters []int | |
| diff --git a/pkg/kv/kvserver/client_protectedts_test.go b/pkg/kv/kvserver/client_protectedts_test.go | |
| index 74eb5378c8..da28e02c3e 100644 | |
| --- a/pkg/kv/kvserver/client_protectedts_test.go | |
| +++ b/pkg/kv/kvserver/client_protectedts_test.go | |
| @@ -20,9 +20,8 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptpb" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptstorage" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptverifier" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig/spanconfigptsreader" | |
| "github.com/cockroachdb/cockroach/pkg/sql" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/skip" | |
| @@ -211,15 +210,12 @@ func TestProtectedTimestamps(t *testing.T) { | |
| thresh := thresholdFromTrace(trace) | |
| require.Truef(t, thresh.Less(ptsRec.Timestamp), "threshold: %v, protected %v %q", thresh, ptsRec.Timestamp, trace) | |
| - // Verify that the record did indeed make its way down into KV where the | |
| - // replica can read it from. | |
| - ptsReader := tc.GetFirstStoreFromServer(t, 0).GetStoreConfig().ProtectedTimestampReader | |
| - require.NoError( | |
| - t, | |
| - verifyProtectionTimestampExistsOnSpans( | |
| - ctx, t, tc, ptsReader, ptsRec.Timestamp, ptsRec.DeprecatedSpans, | |
| - ), | |
| - ) | |
| + // Verify that the record indeed did apply as far as the replica is concerned. | |
| + ptv := ptverifier.New(s0.DB(), pts) | |
| + require.NoError(t, ptv.Verify(ctx, ptsRec.ID.GetUUID())) | |
| + ptsRecVerified, err := ptsWithDB.GetRecord(ctx, nil /* txn */, ptsRec.ID.GetUUID()) | |
| + require.NoError(t, err) | |
| + require.True(t, ptsRecVerified.Verified) | |
| // Make a new record that is doomed to fail. | |
| failedRec := ptsRec | |
| @@ -230,15 +226,9 @@ func TestProtectedTimestamps(t *testing.T) { | |
| _, err = ptsWithDB.GetRecord(ctx, nil /* txn */, failedRec.ID.GetUUID()) | |
| require.NoError(t, err) | |
| - // Verify that the record did indeed make its way down into KV where the | |
| - // replica can read it from. We then verify (below) that the failed record | |
| - // does not affect the ability to GC. | |
| - require.NoError( | |
| - t, | |
| - verifyProtectionTimestampExistsOnSpans( | |
| - ctx, t, tc, ptsReader, failedRec.Timestamp, failedRec.DeprecatedSpans, | |
| - ), | |
| - ) | |
| + // Verify that it indeed did fail. | |
| + verifyErr := ptv.Verify(ctx, failedRec.ID.GetUUID()) | |
| + require.Regexp(t, "failed to verify protection", verifyErr) | |
| // Add a new record that is after the old record. | |
| laterRec := ptsRec | |
| @@ -246,12 +236,7 @@ func TestProtectedTimestamps(t *testing.T) { | |
| laterRec.Timestamp = afterWrites | |
| laterRec.Timestamp.Logical = 0 | |
| require.NoError(t, ptsWithDB.Protect(ctx, nil /* txn */, &laterRec)) | |
| - require.NoError( | |
| - t, | |
| - verifyProtectionTimestampExistsOnSpans( | |
| - ctx, t, tc, ptsReader, laterRec.Timestamp, laterRec.DeprecatedSpans, | |
| - ), | |
| - ) | |
| + require.NoError(t, ptv.Verify(ctx, laterRec.ID.GetUUID())) | |
| // Release the record that had succeeded and ensure that GC eventually | |
| // happens up to the protected timestamp of the new record. | |
| @@ -278,38 +263,3 @@ func TestProtectedTimestamps(t *testing.T) { | |
| require.Len(t, state.Records, 0) | |
| require.Equal(t, int(state.NumRecords), len(state.Records)) | |
| } | |
| - | |
| -// verifyProtectionTimestampExistsOnSpans refreshes the PTS state in KV and | |
| -// ensures a protection at the given protectionTimestamp exists for all the | |
| -// supplied spans. | |
| -func verifyProtectionTimestampExistsOnSpans( | |
| - ctx context.Context, | |
| - t *testing.T, | |
| - tc *testcluster.TestCluster, | |
| - ptsReader spanconfig.ProtectedTSReader, | |
| - protectionTimestamp hlc.Timestamp, | |
| - spans roachpb.Spans, | |
| -) error { | |
| - if err := spanconfigptsreader.TestingRefreshPTSState( | |
| - ctx, t, ptsReader, tc.Server(0).Clock().Now(), | |
| - ); err != nil { | |
| - return err | |
| - } | |
| - for _, sp := range spans { | |
| - timestamps, _, err := ptsReader.GetProtectionTimestamps(ctx, sp) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - found := false | |
| - for _, ts := range timestamps { | |
| - if ts.Equal(protectionTimestamp) { | |
| - found = true | |
| - break | |
| - } | |
| - } | |
| - if !found { | |
| - return errors.Newf("protection timestamp %s does not exist on span %s", protectionTimestamp, sp) | |
| - } | |
| - } | |
| - return nil | |
| -} | |
| diff --git a/pkg/kv/kvserver/client_replica_circuit_breaker_bench_test.go b/pkg/kv/kvserver/client_replica_circuit_breaker_bench_test.go | |
| new file mode 100644 | |
| index 0000000000..686a6a6c55 | |
| --- /dev/null | |
| +++ b/pkg/kv/kvserver/client_replica_circuit_breaker_bench_test.go | |
| @@ -0,0 +1,124 @@ | |
| +// Copyright 2022 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package kvserver_test | |
| + | |
| +import ( | |
| + "context" | |
| + "fmt" | |
| + "math/rand" | |
| + "strconv" | |
| + "sync" | |
| + "testing" | |
| + | |
| + "github.com/cockroachdb/cockroach/pkg/base" | |
| + "github.com/cockroachdb/cockroach/pkg/keys" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver" | |
| + "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| + "github.com/cockroachdb/cockroach/pkg/testutils/testcluster" | |
| + "github.com/cockroachdb/cockroach/pkg/util/encoding" | |
| + "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| + "github.com/cockroachdb/cockroach/pkg/util/log" | |
| + "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| + "github.com/stretchr/testify/require" | |
| +) | |
| + | |
| +type replicaCircuitBreakerBench struct { | |
| + *testcluster.TestCluster | |
| + pool *sync.Pool // *BatchRequest | |
| +} | |
| + | |
| +func (tc *replicaCircuitBreakerBench) repl(b *testing.B) *kvserver.Replica { | |
| + return tc.GetFirstStoreFromServer(b, 0).LookupReplica(keys.MustAddr(tc.ScratchRange(b))) | |
| +} | |
| + | |
| +func setupCircuitBreakerReplicaBench( | |
| + b *testing.B, breakerEnabled bool, cs string, | |
| +) (*replicaCircuitBreakerBench, *stop.Stopper) { | |
| + b.Helper() | |
| + | |
| + var numShards int | |
| + { | |
| + _, err := fmt.Sscanf(cs, "mutexmap-%d", &numShards) | |
| + require.NoError(b, err) | |
| + } | |
| + sFn := func() kvserver.CancelStorage { return &kvserver.MapCancelStorage{NumShards: numShards} } | |
| + | |
| + var knobs kvserver.StoreTestingKnobs | |
| + knobs.CancelStorageFactory = sFn | |
| + | |
| + var args base.TestClusterArgs | |
| + args.ServerArgs.Knobs.Store = &knobs | |
| + tc := testcluster.StartTestCluster(b, 1, args) | |
| + | |
| + stmt := `SET CLUSTER SETTING kv.replica_circuit_breaker.slow_replication_threshold = '1000s'` | |
| + if !breakerEnabled { | |
| + stmt = `SET CLUSTER SETTING kv.replica_circuit_breaker.slow_replication_threshold = '0s'` | |
| + } | |
| + _, err := tc.ServerConn(0).Exec(stmt) | |
| + require.NoError(b, err) | |
| + wtc := &replicaCircuitBreakerBench{ | |
| + TestCluster: tc, | |
| + } | |
| + wtc.pool = &sync.Pool{ | |
| + New: func() interface{} { | |
| + repl := wtc.repl(b) | |
| + var ba roachpb.BatchRequest | |
| + ba.RangeID = repl.RangeID | |
| + ba.Timestamp = repl.Clock().NowAsClockTimestamp().ToTimestamp() | |
| + var k roachpb.Key | |
| + k = append(k, repl.Desc().StartKey.AsRawKey()...) | |
| + k = encoding.EncodeUint64Ascending(k, uint64(rand.Intn(1000))) | |
| + ba.Add(roachpb.NewGet(k, false)) | |
| + return &ba | |
| + }, | |
| + } | |
| + return wtc, tc.Stopper() | |
| +} | |
| + | |
| +func BenchmarkReplicaCircuitBreakerSendOverhead(b *testing.B) { | |
| + defer leaktest.AfterTest(b)() | |
| + defer log.Scope(b).Close(b) | |
| + ctx := context.Background() | |
| + | |
| + for _, enabled := range []bool{false, true} { | |
| + b.Run("enabled="+strconv.FormatBool(enabled), func(b *testing.B) { | |
| + dss := []string{ | |
| + "mutexmap-1", "mutexmap-2", "mutexmap-4", "mutexmap-8", "mutexmap-12", "mutexmap-16", | |
| + "mutexmap-20", "mutexmap-24", "mutexmap-32", "mutexmap-64", | |
| + } | |
| + if !enabled { | |
| + dss = dss[:1] | |
| + } | |
| + | |
| + for _, ds := range dss { | |
| + b.Run(ds, func(b *testing.B) { | |
| + b.ReportAllocs() | |
| + tc, stopper := setupCircuitBreakerReplicaBench(b, enabled, ds) | |
| + defer stopper.Stop(ctx) | |
| + | |
| + repl := tc.repl(b) | |
| + | |
| + b.ResetTimer() | |
| + b.RunParallel(func(pb *testing.PB) { | |
| + for pb.Next() { | |
| + ba := tc.pool.Get().(*roachpb.BatchRequest) | |
| + _, err := repl.Send(ctx, *ba) | |
| + tc.pool.Put(ba) | |
| + if err != nil { | |
| + b.Fatal(err) | |
| + } | |
| + } | |
| + }) | |
| + }) | |
| + } | |
| + }) | |
| + } | |
| +} | |
| diff --git a/pkg/kv/kvserver/client_replica_circuit_breaker_test.go b/pkg/kv/kvserver/client_replica_circuit_breaker_test.go | |
| index de63960736..0293834831 100644 | |
| --- a/pkg/kv/kvserver/client_replica_circuit_breaker_test.go | |
| +++ b/pkg/kv/kvserver/client_replica_circuit_breaker_test.go | |
| @@ -36,7 +36,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| - "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| "github.com/cockroachdb/errors" | |
| "github.com/stretchr/testify/assert" | |
| "github.com/stretchr/testify/require" | |
| @@ -83,13 +82,12 @@ func TestReplicaCircuitBreaker_LeaseholderTripped(t *testing.T) { | |
| defer log.Scope(t).Close(t) | |
| tc := setupCircuitBreakerTest(t) | |
| defer tc.Stopper().Stop(context.Background()) | |
| - k := tc.ScratchRange(t) | |
| // Get lease on n1. | |
| require.NoError(t, tc.Write(n1)) | |
| // Disable the probe so that when the breaker trips, it stays tripped. | |
| tc.SetProbeEnabled(n1, false) | |
| - tc.TripBreaker(n1) | |
| + tc.Report(n1, errors.New("boom")) | |
| s1 := tc.GetFirstStoreFromServer(t, n1) | |
| s2 := tc.GetFirstStoreFromServer(t, n2) | |
| @@ -99,10 +97,10 @@ func TestReplicaCircuitBreaker_LeaseholderTripped(t *testing.T) { | |
| require.Zero(t, s2.Metrics().ReplicaCircuitBreakerCurTripped.Value()) | |
| require.Zero(t, s2.Metrics().ReplicaCircuitBreakerCumTripped.Count()) | |
| - // n1 can still serve reads despite the breaker having tripped, as there is a | |
| - // valid lease and no poisoned latches prevent the read. | |
| - require.NoError(t, tc.Read(n1)) | |
| - require.NoError(t, tc.FollowerRead(n1)) | |
| + // n1 could theoretically still serve reads (there is a valid lease | |
| + // and none of the latches are taken), but since it is hard to determine | |
| + // that upfront we currently fail all reads as well. | |
| + tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| tc.RequireIsBreakerOpen(t, tc.Write(n1)) | |
| // When we go through the KV client stack, we still get the breaker error | |
| @@ -110,29 +108,23 @@ func TestReplicaCircuitBreaker_LeaseholderTripped(t *testing.T) { | |
| tc.RequireIsBreakerOpen(t, tc.WriteDS(n1)) | |
| tc.RequireIsBreakerOpen(t, tc.WriteDS(n2)) | |
| - // Can't transfer the lease away while breaker is tripped. (This would be | |
| - // a bad idea, since n1 would stop serving strong reads, thus making the | |
| - // outage worse). | |
| - tc.RequireIsBreakerOpen(t, | |
| - tc.TransferRangeLease(tc.LookupRangeOrFatal(t, k), tc.Target(n2)), | |
| - ) | |
| - | |
| // n2 does not have the lease so all it does is redirect to the leaseholder | |
| - // n1, but it can serve follower reads. | |
| + // n1. | |
| tc.RequireIsNotLeaseholderError(t, tc.Read(n2)) | |
| - require.NoError(t, tc.FollowerRead(n2)) | |
| tc.RequireIsNotLeaseholderError(t, tc.Write(n2)) | |
| // Enable the probe. Even a read should trigger the probe | |
| // and within due time the breaker should heal. | |
| tc.SetProbeEnabled(n1, true) | |
| - require.NoError(t, tc.Read(n1)) // this always worked | |
| - // Writes heal soon. | |
| + tc.UntripsSoon(t, tc.Read, n1) | |
| + // Same behavior on writes. | |
| + tc.Report(n1, errors.New("boom again")) | |
| tc.UntripsSoon(t, tc.Write, n1) | |
| - // Currently tripped drops back to zero, all-time remains at one. | |
| + // Currently tripped drops back to zero, all-time is two (since we tripped | |
| + // it twice) | |
| require.EqualValues(t, 0, s1.Metrics().ReplicaCircuitBreakerCurTripped.Value()) | |
| - require.EqualValues(t, 1, s1.Metrics().ReplicaCircuitBreakerCumTripped.Count()) | |
| + require.EqualValues(t, 2, s1.Metrics().ReplicaCircuitBreakerCumTripped.Count()) | |
| // s2 wasn't affected by any breaker events. | |
| require.Zero(t, s2.Metrics().ReplicaCircuitBreakerCurTripped.Value()) | |
| require.Zero(t, s2.Metrics().ReplicaCircuitBreakerCumTripped.Count()) | |
| @@ -142,9 +134,9 @@ func TestReplicaCircuitBreaker_LeaseholderTripped(t *testing.T) { | |
| // breaker on follower n2. Before the breaker is tripped, we see | |
| // NotLeaseholderError. When it's tripped, those are supplanted by the breaker | |
| // errors. Once we allow the breaker to probe, the breaker untrips. In | |
| -// particular, this tests that the probe can succeed even when run on a follower | |
| -// (which would not be true if it required the local Replica to execute an | |
| -// operation that requires the lease). | |
| +// particular, this tests that the probe can succeed even when run on a | |
| +// follower (which would not be true if it required the local Replica to | |
| +// execute an operation that requires the lease). | |
| func TestReplicaCircuitBreaker_FollowerTripped(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| @@ -155,7 +147,7 @@ func TestReplicaCircuitBreaker_FollowerTripped(t *testing.T) { | |
| require.NoError(t, tc.Write(n1)) | |
| // Disable the probe on n2 so that when the breaker trips, it stays tripped. | |
| tc.SetProbeEnabled(n2, false) | |
| - tc.TripBreaker(n2) | |
| + tc.Report(n2, errors.New("boom")) | |
| // We didn't trip the leaseholder n1, so it is unaffected. | |
| require.NoError(t, tc.Read(n1)) | |
| @@ -166,21 +158,31 @@ func TestReplicaCircuitBreaker_FollowerTripped(t *testing.T) { | |
| // time of writing it would propagate it. | |
| require.NoError(t, tc.WriteDS(n1)) | |
| - tc.RequireIsNotLeaseholderError(t, tc.Read(n2)) | |
| - tc.RequireIsNotLeaseholderError(t, tc.Write(n2)) | |
| - require.NoError(t, tc.FollowerRead(n2)) | |
| + tc.RequireIsBreakerOpen(t, tc.Read(n2)) | |
| + tc.RequireIsBreakerOpen(t, tc.Write(n2)) | |
| - // Enable the probe again. n2 should untrip soon. | |
| + // Enable the probe. Even a read should trigger the probe | |
| + // and within due time the breaker should heal, giving us | |
| + // NotLeaseholderErrors again. | |
| + // | |
| + // TODO(tbg): this test would be more meaningful with follower reads. They | |
| + // should succeed when the breaker is open and fail if the breaker is | |
| + // tripped. However knowing that the circuit breaker check sits at the top | |
| + // of Replica.sendWithRangeID, it's clear that it won't make a difference. | |
| tc.SetProbeEnabled(n2, true) | |
| - tc.RequireIsNotLeaseholderError(t, tc.Read(n2)) | |
| - tc.RequireIsNotLeaseholderError(t, tc.Write(n2)) | |
| testutils.SucceedsSoon(t, func() error { | |
| - // NB: this is slightly contrived - the mere act of accessing Err() is what | |
| - // triggers the probe! Regular requests on the replica wouldn'd do that, | |
| - // since we're intentionally preferring a NotLeaseholderError over a breaker | |
| - // error (and thus aren't ever accessing the breaker when we can't serve the | |
| - // request). | |
| - return tc.repls[n2].Breaker().Signal().Err() | |
| + if err := tc.Read(n2); !errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) { | |
| + return err | |
| + } | |
| + return nil | |
| + }) | |
| + // Same behavior on writes. | |
| + tc.Report(n2, errors.New("boom again")) | |
| + testutils.SucceedsSoon(t, func() error { | |
| + if err := tc.Write(n2); !errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) { | |
| + return err | |
| + } | |
| + return nil | |
| }) | |
| } | |
| @@ -200,45 +202,34 @@ func TestReplicaCircuitBreaker_LeaselessTripped(t *testing.T) { | |
| // disabled. | |
| require.NoError(t, tc.Write(n1)) | |
| tc.SetProbeEnabled(n1, false) | |
| - tc.TripBreaker(n1) | |
| - resumeHeartbeats := tc.ExpireAllLeasesAndN1LivenessRecord(t, pauseHeartbeats) | |
| - | |
| - // On n1, run into the circuit breaker when requesting lease. We have to | |
| - // resume heartbeats for this to not time out, as requesting the new lease | |
| - // entails doing liveness checks which can't succeed if nobody is | |
| - // heartbeating, and we'd get stuck in liveness before reaching the circuit | |
| - // breaker. (In other words, replica circuit breaking doesn't fail-fast | |
| - // requests reliably if liveness is unavailable; this is tracked in #74616). | |
| - // We don't attempt to acquire a lease on n2 since it would try and succeed | |
| - // (except the test harness categorically prevents n2 from getting a lease, | |
| - // injecting an error). | |
| - resumeHeartbeats() | |
| - testutils.SucceedsSoon(t, func() error { | |
| - err := tc.Read(n1) | |
| - if errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) { | |
| - // Retriable errors can occur when manipulating the liveness record in | |
| - // preparation for requesting a lease, such as: | |
| - // | |
| - // [NotLeaseHolderError] failed to manipulate liveness record: heartbeat | |
| - // failed on epoch increment; r45: replica (n1,s1):1 not lease holder; | |
| - // current lease is repl=(n1,s1):1 seq=1 start=0,0 epo=1 pro=[...] | |
| - return err | |
| - } | |
| - tc.RequireIsBreakerOpen(t, err) | |
| - tc.RequireIsBreakerOpen(t, tc.Write(n1)) | |
| - return nil | |
| - }) | |
| + tc.Report(n1, errors.New("boom")) | |
| + resumeHeartbeats := tc.ExpireAllLeases(t, pauseHeartbeats) | |
| - // Can still perform follower reads on both nodes, as this does not rely on | |
| - // the lease and does not consult the breaker. | |
| - require.NoError(t, tc.FollowerRead(n1)) | |
| - require.NoError(t, tc.FollowerRead(n2)) | |
| + // n2 (not n1) will return a NotLeaseholderError. This may be surprising - | |
| + // why isn't it trying and succeeding to acquire a lease - but it does | |
| + // not do that because it sees that the new leaseholder (n2) is not live | |
| + // itself. We'll revisit this after re-enabling liveness later in the test. | |
| + { | |
| + err := tc.Read(n2) | |
| + // At time of writing: not incrementing epoch on n1 because next | |
| + // leaseholder (n2) not live. | |
| + t.Log(err) | |
| + tc.RequireIsNotLeaseholderError(t, err) | |
| + // Same behavior for write on n2. | |
| + tc.RequireIsNotLeaseholderError(t, tc.Write(n2)) | |
| + } | |
| + // On n1, run into the circuit breaker when requesting lease. | |
| + { | |
| + tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| + tc.RequireIsBreakerOpen(t, tc.Write(n1)) | |
| + } | |
| // Let the breaker heal and things should go back to normal. This is not a | |
| // trivial thing to hold, as the probe needs to go through for this, and if | |
| // we're not careful, the probe itself is held up by the breaker as well, or | |
| // the probe will try to acquire a lease (which we're currently careful to | |
| // avoid). | |
| + resumeHeartbeats() | |
| tc.SetProbeEnabled(n1, true) | |
| tc.UntripsSoon(t, tc.Read, n1) | |
| tc.UntripsSoon(t, tc.Write, n1) | |
| @@ -263,33 +254,21 @@ func TestReplicaCircuitBreaker_Leaseholder_QuorumLoss(t *testing.T) { | |
| tc.StopServer(n2) // lose quorum | |
| // We didn't lose the liveness range (which is only on n1). | |
| - tc.HeartbeatNodeLiveness(t, n1) | |
| - | |
| - // Read still works, as we have valid lease and no poisoned latch | |
| - // underneath. | |
| - require.NoError(t, tc.Read(n1)) | |
| + require.NoError(t, tc.Server(n1).HeartbeatNodeLiveness()) | |
| tc.SetSlowThreshold(10 * time.Millisecond) | |
| { | |
| err := tc.Write(n1) | |
| var ae *roachpb.AmbiguousResultError | |
| require.True(t, errors.As(err, &ae), "%+v", err) | |
| t.Log(err) | |
| - tc.RequireIsBreakerOpen(t, err) | |
| - } | |
| - // We still have a valid lease, but now the above write is holding a poisoned | |
| - // latch (this is true despite the write itself having returned already). | |
| - // However, can still serve follower reads because those don't check latches | |
| - // (nor do they need the lease, though there is a valid one in this case). | |
| - { | |
| - tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| - require.NoError(t, tc.FollowerRead(n1)) | |
| } | |
| + tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| // Bring n2 back and service should be restored. | |
| tc.SetSlowThreshold(0) // reset | |
| require.NoError(t, tc.RestartServer(n2)) | |
| - tc.UntripsSoon(t, tc.Write, n1) // poisoned latch goes away | |
| - require.NoError(t, tc.Read(n1)) | |
| + tc.UntripsSoon(t, tc.Read, n1) | |
| + require.NoError(t, tc.Write(n1)) | |
| } | |
| // In this test, the range is on n1 and n2 and we place the lease on n2 and | |
| @@ -308,13 +287,12 @@ func TestReplicaCircuitBreaker_Follower_QuorumLoss(t *testing.T) { | |
| // Get lease to n2 so that we can lose it without taking down the system ranges. | |
| desc := tc.LookupRangeOrFatal(t, tc.ScratchRange(t)) | |
| tc.TransferRangeLeaseOrFatal(t, desc, tc.Target(n2)) | |
| - resumeHeartbeats := tc.ExpireAllLeasesAndN1LivenessRecord(t, keepHeartbeats) | |
| + resumeHeartbeats := tc.ExpireAllLeases(t, keepHeartbeats) | |
| tc.StopServer(n2) // lose quorum and leaseholder | |
| resumeHeartbeats() | |
| // We didn't lose the liveness range (which is only on n1). | |
| - tc.HeartbeatNodeLiveness(t, n1) | |
| - | |
| + require.NoError(t, tc.Server(n1).HeartbeatNodeLiveness()) | |
| tc.SetSlowThreshold(10 * time.Millisecond) | |
| tc.RequireIsBreakerOpen(t, tc.Write(n1)) | |
| tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| @@ -322,8 +300,8 @@ func TestReplicaCircuitBreaker_Follower_QuorumLoss(t *testing.T) { | |
| // Bring n2 back and service should be restored. | |
| tc.SetSlowThreshold(0) // reset | |
| require.NoError(t, tc.RestartServer(n2)) | |
| - tc.UntripsSoon(t, tc.Write, n1) | |
| - require.NoError(t, tc.Read(n1)) | |
| + tc.UntripsSoon(t, tc.Read, n1) | |
| + require.NoError(t, tc.Write(n1)) | |
| } | |
| // This test is skipped but documents that the current circuit breakers cannot | |
| @@ -375,7 +353,7 @@ func TestReplicaCircuitBreaker_Liveness_QuorumLoss(t *testing.T) { | |
| // Expire all leases. We also pause all heartbeats but that doesn't really | |
| // matter since the liveness range is unavailable anyway. | |
| - resume := tc.ExpireAllLeasesAndN1LivenessRecord(t, pauseHeartbeats) | |
| + resume := tc.ExpireAllLeases(t, pauseHeartbeats) | |
| defer resume() | |
| // Since there isn't a lease, and the liveness range is down, the circuit | |
| @@ -395,11 +373,6 @@ func TestReplicaCircuitBreaker_Liveness_QuorumLoss(t *testing.T) { | |
| } | |
| type dummyStream struct { | |
| - name string | |
| - t interface { | |
| - Helper() | |
| - Logf(string, ...interface{}) | |
| - } | |
| ctx context.Context | |
| recv chan *roachpb.RangeFeedEvent | |
| } | |
| @@ -409,8 +382,7 @@ func (s *dummyStream) Context() context.Context { | |
| } | |
| func (s *dummyStream) Send(ev *roachpb.RangeFeedEvent) error { | |
| - if ev.Val == nil && ev.Error == nil { | |
| - s.t.Logf("%s: ignoring event: %v", s.name, ev) | |
| + if ev.Val == nil { | |
| return nil | |
| } | |
| select { | |
| @@ -438,44 +410,39 @@ func TestReplicaCircuitBreaker_RangeFeed(t *testing.T) { | |
| args := &roachpb.RangeFeedRequest{ | |
| Span: roachpb.Span{Key: desc.StartKey.AsRawKey(), EndKey: desc.EndKey.AsRawKey()}, | |
| } | |
| - | |
| - ctx, cancel := context.WithCancel(ctx) | |
| + // This test shouldn't take in excess of 45s even under the worst of conditions. | |
| + ctx, cancel := context.WithTimeout(ctx, testutils.DefaultSucceedsSoonDuration) | |
| defer cancel() | |
| - stream1 := &dummyStream{t: t, ctx: ctx, name: "rangefeed1", recv: make(chan *roachpb.RangeFeedEvent)} | |
| + stream1 := &dummyStream{ctx: ctx, recv: make(chan *roachpb.RangeFeedEvent)} | |
| require.NoError(t, tc.Stopper().RunAsyncTask(ctx, "stream1", func(ctx context.Context) { | |
| err := tc.repls[0].RangeFeed(args, stream1).GoError() | |
| if ctx.Err() != nil { | |
| - return // main goroutine stopping | |
| + return | |
| } | |
| assert.NoError(t, err) // avoid Fatal on goroutine | |
| })) | |
| - readOneVal := func(ctx context.Context, stream *dummyStream, timeout time.Duration) error { | |
| + readOneVal := func(t *testing.T, stream *dummyStream) { | |
| for { | |
| + var done bool | |
| select { | |
| - case <-time.After(timeout): | |
| - return errors.Errorf("%s: read timed out after %.2fs", stream.name, timeout.Seconds()) | |
| case <-ctx.Done(): | |
| - return ctx.Err() | |
| + t.Fatal(ctx.Err()) | |
| case ev := <-stream.recv: | |
| - if ev.Error != nil { | |
| - return ev.Error.Error.GoError() | |
| - } | |
| - if ev.Val != nil { | |
| - t.Logf("%s: %s", stream.name, ev) | |
| - return nil | |
| - } | |
| + t.Log(ev) | |
| + done = true | |
| + } | |
| + if done { | |
| + break | |
| } | |
| } | |
| } | |
| - testutils.SucceedsSoon(t, func() error { | |
| - require.NoError(t, tc.Write(n1)) | |
| - return readOneVal(ctx, stream1, time.Millisecond) | |
| - }) | |
| + require.NoError(t, tc.Write(n1)) | |
| + readOneVal(t, stream1) | |
| // NB: keep heartbeats because we're not trying to lose the liveness range. | |
| - undo := tc.ExpireAllLeasesAndN1LivenessRecord(t, keepHeartbeats) | |
| + undo := tc.ExpireAllLeases(t, keepHeartbeats) | |
| undo() | |
| tc.SetSlowThreshold(10 * time.Millisecond) | |
| tc.StopServer(n2) | |
| @@ -483,11 +450,11 @@ func TestReplicaCircuitBreaker_RangeFeed(t *testing.T) { | |
| // Start another stream during the "outage" to make sure it isn't rejected by | |
| // the breaker. | |
| - stream2 := &dummyStream{t: t, ctx: ctx, name: "rangefeed2", recv: make(chan *roachpb.RangeFeedEvent)} | |
| + stream2 := &dummyStream{ctx: ctx, recv: make(chan *roachpb.RangeFeedEvent)} | |
| require.NoError(t, tc.Stopper().RunAsyncTask(ctx, "stream2", func(ctx context.Context) { | |
| err := tc.repls[0].RangeFeed(args, stream2).GoError() | |
| if ctx.Err() != nil { | |
| - return // main goroutine stopping | |
| + return | |
| } | |
| assert.NoError(t, err) // avoid Fatal on goroutine | |
| })) | |
| @@ -496,13 +463,8 @@ func TestReplicaCircuitBreaker_RangeFeed(t *testing.T) { | |
| require.NoError(t, tc.RestartServer(n2)) | |
| tc.UntripsSoon(t, tc.Write, n1) | |
| - require.NoError(t, readOneVal(ctx, stream1, testutils.DefaultSucceedsSoonDuration)) | |
| - // For the stream that started mid-way through the outage, we expect it to | |
| - // return a circuit breaker error, but in theory it could also never have | |
| - // tried to acquire a lease, in which case it might return a value as well. | |
| - if err := readOneVal(ctx, stream2, testutils.DefaultSucceedsSoonDuration); err != nil { | |
| - tc.RequireIsBreakerOpen(t, err) | |
| - } | |
| + readOneVal(t, stream1) | |
| + readOneVal(t, stream2) | |
| } | |
| func TestReplicaCircuitBreaker_ExemptRequests(t *testing.T) { | |
| @@ -515,7 +477,7 @@ func TestReplicaCircuitBreaker_ExemptRequests(t *testing.T) { | |
| // disabled, i.e. it will stay tripped. | |
| require.NoError(t, tc.Write(n1)) | |
| tc.SetProbeEnabled(n1, false) | |
| - tc.TripBreaker(n1) | |
| + tc.Report(n1, errors.New("boom")) | |
| exemptRequests := []func() roachpb.Request{ | |
| func() roachpb.Request { return &roachpb.ExportRequest{ReturnSST: true} }, | |
| @@ -548,41 +510,31 @@ func TestReplicaCircuitBreaker_ExemptRequests(t *testing.T) { | |
| for _, reqFn := range exemptRequests { | |
| req := reqFn() | |
| - tc.Run(t, fmt.Sprintf("with-existing-lease/%s", req.Method()), func(t *testing.T) { | |
| + t.Run(fmt.Sprintf("with-existing-lease/%s", req.Method()), func(t *testing.T) { | |
| require.NoError(t, tc.Send(n1, req)) | |
| }) | |
| } | |
| for _, reqFn := range exemptRequests { | |
| req := reqFn() | |
| - tc.Run(t, fmt.Sprintf("with-acquire-lease/%s", req.Method()), func(t *testing.T) { | |
| - resumeHeartbeats := tc.ExpireAllLeasesAndN1LivenessRecord(t, pauseHeartbeats) | |
| + t.Run(fmt.Sprintf("with-acquire-lease/%s", req.Method()), func(t *testing.T) { | |
| + resumeHeartbeats := tc.ExpireAllLeases(t, pauseHeartbeats) | |
| resumeHeartbeats() // intentionally resume right now so that lease can be acquired | |
| - // NB: when looking into the traces here, we sometimes see - as expected - | |
| - // that when the request tries to acquire a lease, the breaker is still | |
| - // tripped. That's why there is a retry loop here. | |
| - testutils.SucceedsSoon(t, func() error { | |
| - err := tc.Send(n1, req) | |
| - if errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) { | |
| - return err | |
| - } | |
| - require.NoError(t, err) | |
| - return nil | |
| - }) | |
| + require.NoError(t, tc.Send(n1, req)) | |
| }) | |
| } | |
| - resumeHeartbeats := tc.ExpireAllLeasesAndN1LivenessRecord(t, pauseHeartbeats) | |
| + resumeHeartbeats := tc.ExpireAllLeases(t, pauseHeartbeats) | |
| + defer resumeHeartbeats() // can't acquire leases until test ends | |
| for _, reqFn := range exemptRequests { | |
| req := reqFn() | |
| - tc.Run(t, fmt.Sprintf("with-unavailable-lease/%s", req.Method()), func(t *testing.T) { | |
| - if m := req.Method(); m == roachpb.Probe { | |
| - // Probe does not require the lease, and is the most-tested of the bunch | |
| - // already. We don't have to test it again here, which would require undue | |
| - // amounts of special-casing below. | |
| - skip.IgnoreLintf(t, "subtest does not apply to %s", m) | |
| - } | |
| - | |
| + if req.Method() == roachpb.Probe { | |
| + // Probe does not require the lease, and is the most-tested of the bunch | |
| + // already. We don't have to test it again here, which would require undue | |
| + // amounts of special-casing below. | |
| + continue | |
| + } | |
| + t.Run(fmt.Sprintf("with-unavailable-lease/%s", req.Method()), func(t *testing.T) { | |
| ctx, cancel := context.WithTimeout(context.Background(), 3*time.Millisecond) | |
| defer cancel() | |
| const maxWait = 5 * time.Second | |
| @@ -596,46 +548,6 @@ func TestReplicaCircuitBreaker_ExemptRequests(t *testing.T) { | |
| require.Less(t, timeutil.Since(tBegin), maxWait) | |
| }) | |
| } | |
| - | |
| - // Restore the breaker via the probe. | |
| - resumeHeartbeats() | |
| - tc.SetProbeEnabled(n1, true) | |
| - tc.UntripsSoon(t, tc.Write, n1) | |
| - | |
| - // Lose quorum (liveness stays intact). | |
| - tc.SetSlowThreshold(10 * time.Millisecond) | |
| - tc.StopServer(n2) | |
| - // Let the breaker trip. This leaves a poisoned latch behind that at least some of | |
| - // the requests will interact with. | |
| - tc.RequireIsBreakerOpen(t, tc.Write(n1)) | |
| - tc.RequireIsBreakerOpen(t, tc.Read(n1)) | |
| - | |
| - for _, reqFn := range exemptRequests { | |
| - req := reqFn() | |
| - tc.Run(t, fmt.Sprintf("with-poisoned-latch/%s", req.Method()), func(t *testing.T) { | |
| - if m := req.Method(); m == roachpb.GC { | |
| - // GC without GCKeys acquires no latches and is a pure read. If we want | |
| - // to put a key in there, we need to pick the right timestamp (since you | |
| - // can't GC a live key); it's all rather annoying and not worth it. In | |
| - // the long run, we also completely want to avoid acquiring latches for | |
| - // this request (since it should only mutate keyspace that has since | |
| - // fallen under the GCThreshold), so avoid cooking up anything special | |
| - // here. | |
| - skip.IgnoreLintf(t, "subtest does not apply to %s", m) | |
| - } | |
| - ctx, cancel := context.WithTimeout(context.Background(), 3*time.Millisecond) | |
| - defer cancel() | |
| - const maxWait = 5 * time.Second | |
| - tBegin := timeutil.Now() | |
| - err := tc.SendCtx(ctx, n1, req) | |
| - t.Log(err) | |
| - require.Error(t, err) | |
| - require.Error(t, ctx.Err()) | |
| - // Make sure we didn't run into the "long" timeout inside of SendCtx but | |
| - // actually terminated as a result of our ctx cancelling. | |
| - require.Less(t, timeutil.Since(tBegin), maxWait) | |
| - }) | |
| - } | |
| } | |
| // Test infrastructure below. | |
| @@ -661,18 +573,19 @@ func makeBreakerToggleable(b *circuit.Breaker) (setProbeEnabled func(bool)) { | |
| } | |
| } | |
| +type replWithKnob struct { | |
| + *kvserver.Replica | |
| + setProbeEnabled func(bool) | |
| +} | |
| + | |
| type circuitBreakerTest struct { | |
| - t decoT | |
| *testcluster.TestCluster | |
| slowThresh *atomic.Value // time.Duration | |
| ManualClock *hlc.HybridManualClock | |
| repls []replWithKnob // 0 -> repl on Servers[0], etc | |
| - | |
| - seq int | |
| } | |
| func setupCircuitBreakerTest(t *testing.T) *circuitBreakerTest { | |
| - skip.UnderStressRace(t) | |
| manualClock := hlc.NewHybridManualClock() | |
| var rangeID int64 // atomic | |
| slowThresh := &atomic.Value{} // supports .SetSlowThreshold(x) | |
| @@ -743,8 +656,7 @@ func setupCircuitBreakerTest(t *testing.T) *circuitBreakerTest { | |
| k := tc.ScratchRange(t) | |
| atomic.StoreInt64(&rangeID, int64(tc.LookupRangeOrFatal(t, k).RangeID)) | |
| - tc.AddVotersOrFatal(t, k, tc.Target(n2)) | |
| - require.NoError(t, tc.WaitForVoters(k, tc.Target(n2))) | |
| + tc.AddVotersOrFatal(t, k, tc.Target(1)) | |
| var repls []replWithKnob | |
| for i := range tc.Servers { | |
| @@ -753,7 +665,6 @@ func setupCircuitBreakerTest(t *testing.T) *circuitBreakerTest { | |
| repls = append(repls, replWithKnob{repl, enableProbe}) | |
| } | |
| return &circuitBreakerTest{ | |
| - t: decoT{t}, | |
| TestCluster: tc, | |
| ManualClock: manualClock, | |
| repls: repls, | |
| @@ -761,28 +672,12 @@ func setupCircuitBreakerTest(t *testing.T) *circuitBreakerTest { | |
| } | |
| } | |
| -// Run is a wrapper around t.Run that allows the test harness to print traces | |
| -// using the subtest's *testing.T. | |
| -func (cbt *circuitBreakerTest) Run(t *testing.T, name string, f func(t *testing.T)) { | |
| - t.Helper() | |
| - t.Run(name, func(t *testing.T) { | |
| - t.Helper() | |
| - outerT := cbt.t | |
| - cbt.t = decoT{t} | |
| - defer func() { | |
| - cbt.t = outerT | |
| - }() | |
| - f(t) | |
| - }) | |
| -} | |
| - | |
| func (cbt *circuitBreakerTest) SetProbeEnabled(idx int, to bool) { | |
| cbt.repls[idx].setProbeEnabled(to) | |
| } | |
| -func (cbt *circuitBreakerTest) TripBreaker(idx int) { | |
| - repl := cbt.repls[idx].Replica | |
| - repl.TripBreaker() | |
| +func (cbt *circuitBreakerTest) Report(idx int, err error) { | |
| + cbt.repls[idx].Replica.Breaker().Report(err) | |
| } | |
| func (cbt *circuitBreakerTest) UntripsSoon(t *testing.T, method func(idx int) error, idx int) { | |
| @@ -791,60 +686,31 @@ func (cbt *circuitBreakerTest) UntripsSoon(t *testing.T, method func(idx int) er | |
| t.Helper() | |
| err := method(idx) | |
| // All errors coming out should be annotated as coming from | |
| - // the circuit breaker. In rare cases, we can also see a | |
| - // NotLeaseholderError such as this one: | |
| - // [NotLeaseHolderError] failed to manipulate liveness record: heartbeat | |
| - // failed on epoch increment; r45: replica (n1,s1):1 not lease holder; | |
| - // current lease is repl=(n1,s1):1 seq=1 start=0,0 epo=1 pro=[...] | |
| - if err != nil && | |
| - !errors.Is(err, circuit.ErrBreakerOpen) && | |
| - !errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) { | |
| - | |
| + // the circuit breaker. | |
| + if err != nil && !errors.Is(err, circuit.ErrBreakerOpen) { | |
| t.Fatalf("saw unexpected error %+v", err) | |
| } | |
| return err | |
| }) | |
| } | |
| -func (cbt *circuitBreakerTest) ExpireAllLeasesAndN1LivenessRecord( | |
| - t *testing.T, pauseHeartbeats bool, | |
| -) (undo func()) { | |
| +func (cbt *circuitBreakerTest) ExpireAllLeases(t *testing.T, pauseHeartbeats bool) (undo func()) { | |
| t.Helper() | |
| + var maxWT int64 | |
| var fs []func() | |
| - for idx, srv := range cbt.Servers { | |
| + for _, srv := range cbt.Servers { | |
| lv := srv.NodeLiveness().(*liveness.NodeLiveness) | |
| - | |
| if pauseHeartbeats { | |
| undo := lv.PauseAllHeartbeatsForTest() | |
| fs = append(fs, undo) | |
| } | |
| - | |
| self, ok := lv.Self() | |
| require.True(t, ok) | |
| - | |
| - cbt.ManualClock.Forward(self.Expiration.WallTime) | |
| - if idx == n1 { | |
| - // Invalidate n1's liveness record, to make sure that ranges on n1 need | |
| - // to acquire a new lease (vs waiting for a heartbeat to the liveness | |
| - // record resuscitating the old one). | |
| - // | |
| - // Needing to do this is the reason for special-casing this entire method | |
| - // around n1; if we stop heartbeats for both nodes, they can't increment | |
| - // each others liveness records: if a node's liveness is paused, it doesn't | |
| - // allow incrementing records neither. (This is silly). | |
| - lv2 := cbt.Server(n2).NodeLiveness().(*liveness.NodeLiveness) | |
| - testutils.SucceedsSoon(t, func() error { | |
| - self, ok := lv.Self() | |
| - require.True(t, ok) | |
| - if self.IsLive(cbt.Server(n2).Clock().Now().GoTime()) { | |
| - // Someone else must have incremented epoch. | |
| - return nil | |
| - } | |
| - return lv2.IncrementEpoch(context.Background(), self) | |
| - }) | |
| + if maxWT < self.Expiration.WallTime { | |
| + maxWT = self.Expiration.WallTime | |
| } | |
| } | |
| - | |
| + cbt.ManualClock.Forward(maxWT + 1) | |
| return func() { | |
| for _, f := range fs { | |
| f() | |
| @@ -853,32 +719,15 @@ func (cbt *circuitBreakerTest) ExpireAllLeasesAndN1LivenessRecord( | |
| } | |
| func (cbt *circuitBreakerTest) Send(idx int, req roachpb.Request) error { | |
| - cbt.t.Helper() | |
| return cbt.SendCtx(context.Background(), idx, req) | |
| -} | |
| -func (cbt *circuitBreakerTest) SendCtx(ctx context.Context, idx int, req roachpb.Request) error { | |
| - return cbt.SendCtxTS(ctx, idx, req, cbt.repls[idx].Clock().Now()) | |
| } | |
| -func (cbt *circuitBreakerTest) SendCtxTS( | |
| - ctx context.Context, idx int, req roachpb.Request, ts hlc.Timestamp, | |
| -) error { | |
| - cbt.t.Helper() | |
| - ctx, finishAndGet := tracing.ContextWithRecordingSpan(ctx, cbt.repls[idx].Tracer, "SendCtx("+req.Method().String()+")") | |
| - defer time.AfterFunc(10*time.Second, func() { | |
| - rec := tracing.SpanFromContext(ctx).GetConfiguredRecording() | |
| - cbt.t.Logf("slow request: %s", rec) | |
| - }).Stop() | |
| - defer func() { | |
| - cbt.t.Helper() | |
| - rec := finishAndGet() | |
| - cbt.t.Logf("%s", rec) | |
| - }() | |
| +func (cbt *circuitBreakerTest) SendCtx(ctx context.Context, idx int, req roachpb.Request) error { | |
| var ba roachpb.BatchRequest | |
| repl := cbt.repls[idx] | |
| ba.RangeID = repl.Desc().RangeID | |
| - ba.Timestamp = ts | |
| + ba.Timestamp = repl.Clock().Now() | |
| ba.Add(req) | |
| if h := req.Header(); len(h.Key) == 0 { | |
| h.Key = repl.Desc().StartKey.AsRawKey() | |
| @@ -902,20 +751,29 @@ func (cbt *circuitBreakerTest) SendCtxTS( | |
| if err := ctx.Err(); err != nil && parCtx.Err() == nil { | |
| pErr = roachpb.NewErrorf("timed out waiting for batch response: %v", pErr) | |
| } | |
| + { | |
| + var err error | |
| + repl.VisitBreakerContexts(func(ctx context.Context) { | |
| + if err == nil && ctx.Value(req) != nil { | |
| + err = errors.Errorf( | |
| + "request %s returned but context still tracked in breaker", req, | |
| + ) | |
| + } | |
| + }) | |
| + if err != nil { | |
| + pErr = roachpb.NewErrorf("%s; after %v", err, pErr) | |
| + } | |
| + } | |
| return pErr.GoError() | |
| } | |
| func (cbt *circuitBreakerTest) WriteDS(idx int) error { | |
| - cbt.t.Helper() | |
| put := roachpb.NewPut(cbt.repls[idx].Desc().StartKey.AsRawKey(), roachpb.MakeValueFromString("hello")) | |
| return cbt.sendViaDistSender(cbt.Servers[idx].DistSender(), put) | |
| } | |
| -func (cbt *circuitBreakerTest) sendViaDistSender( | |
| - ds *kvcoord.DistSender, req roachpb.Request, | |
| -) error { | |
| - cbt.t.Helper() | |
| +func (*circuitBreakerTest) sendViaDistSender(ds *kvcoord.DistSender, req roachpb.Request) error { | |
| var ba roachpb.BatchRequest | |
| ba.Add(req) | |
| ctx, cancel := context.WithTimeout(context.Background(), testutils.DefaultSucceedsSoonDuration) | |
| @@ -932,17 +790,16 @@ func (cbt *circuitBreakerTest) sendViaDistSender( | |
| func (*circuitBreakerTest) RequireIsBreakerOpen(t *testing.T, err error) { | |
| t.Helper() | |
| - t.Log(err) | |
| - // NB: we will see AmbiguousResultError here when proposals are inflight while | |
| - // the breaker trips. These are wrapping errors, so the assertions below will | |
| - // look through them. | |
| + // We also accept an ambiguous result wrapping a breaker error; this occurs | |
| + // when the breaker trips while a write is already inflight. | |
| + if aErr := (*roachpb.AmbiguousResultError)(nil); errors.As(err, &aErr) && aErr.WrappedErr != nil { | |
| + err = aErr.WrappedErr.GoError() | |
| + } | |
| require.True(t, errors.Is(err, circuit.ErrBreakerOpen), "%+v", err) | |
| - require.True(t, errors.HasType(err, (*roachpb.ReplicaUnavailableError)(nil)), "%+v", err) | |
| } | |
| func (*circuitBreakerTest) RequireIsNotLeaseholderError(t *testing.T, err error) { | |
| t.Helper() | |
| - t.Log(err) | |
| ok := errors.HasType(err, (*roachpb.NotLeaseHolderError)(nil)) | |
| require.True(t, ok, "%+v", err) | |
| } | |
| @@ -955,56 +812,13 @@ func (cbt *circuitBreakerTest) SetSlowThreshold(dur time.Duration) { | |
| } | |
| func (cbt *circuitBreakerTest) Write(idx int) error { | |
| - cbt.t.Helper() | |
| repl := cbt.repls[idx] | |
| - cbt.seq++ | |
| - put := roachpb.NewPut( | |
| - repl.Desc().StartKey.AsRawKey(), roachpb.MakeValueFromString(fmt.Sprintf("hello-%d", cbt.seq)), | |
| - ) | |
| + put := roachpb.NewPut(repl.Desc().StartKey.AsRawKey(), roachpb.MakeValueFromString("hello")) | |
| return cbt.Send(idx, put) | |
| } | |
| func (cbt *circuitBreakerTest) Read(idx int) error { | |
| - cbt.t.Helper() | |
| repl := cbt.repls[idx] | |
| get := roachpb.NewGet(repl.Desc().StartKey.AsRawKey(), false /* forUpdate */) | |
| return cbt.Send(idx, get) | |
| } | |
| - | |
| -func (cbt *circuitBreakerTest) FollowerRead(idx int) error { | |
| - cbt.t.Helper() | |
| - repl := cbt.repls[idx] | |
| - get := roachpb.NewGet(repl.Desc().StartKey.AsRawKey(), false /* forUpdate */) | |
| - ctx := context.Background() | |
| - ts := repl.GetClosedTimestamp(ctx) | |
| - return cbt.SendCtxTS(ctx, idx, get, ts) | |
| -} | |
| - | |
| -func (cbt *circuitBreakerTest) HeartbeatNodeLiveness(t *testing.T, idx int) { | |
| - // Retry loop is needed because heartbeat may race with internal heartbeat | |
| - // loop. | |
| - testutils.SucceedsSoon(t, func() error { | |
| - return cbt.Server(idx).HeartbeatNodeLiveness() | |
| - }) | |
| -} | |
| - | |
| -type replWithKnob struct { | |
| - *kvserver.Replica | |
| - setProbeEnabled func(bool) | |
| -} | |
| - | |
| -type logT interface { | |
| - Helper() | |
| - Logf(string, ...interface{}) | |
| -} | |
| - | |
| -type decoT struct { | |
| - logT | |
| -} | |
| - | |
| -func (t *decoT) Logf(format string, args ...interface{}) { | |
| - // It can be difficult to spot the actual failure among all of the | |
| - // traces, so this is a convenient place to annotate the logging | |
| - // (or disable it one-off). | |
| - t.logT.Logf("info:\n"+format, args...) | |
| -} | |
| diff --git a/pkg/kv/kvserver/client_replica_test.go b/pkg/kv/kvserver/client_replica_test.go | |
| index 681c1ba188..a14aa81a3a 100644 | |
| --- a/pkg/kv/kvserver/client_replica_test.go | |
| +++ b/pkg/kv/kvserver/client_replica_test.go | |
| @@ -38,7 +38,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/server" | |
| "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| "github.com/cockroachdb/cockroach/pkg/spanconfig" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig/spanconfigptsreader" | |
| "github.com/cockroachdb/cockroach/pkg/sql" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/bootstrap" | |
| "github.com/cockroachdb/cockroach/pkg/storage" | |
| @@ -3518,12 +3517,10 @@ func TestStrictGCEnforcement(t *testing.T) { | |
| t.Helper() | |
| testutils.SucceedsSoon(t, func() error { | |
| for i := 0; i < tc.NumServers(); i++ { | |
| - ptsReader := tc.GetFirstStoreFromServer(t, 0).GetStoreConfig().ProtectedTimestampReader | |
| - _, asOf, err := ptsReader.GetProtectionTimestamps(ctx, tableSpan) | |
| - if err != nil { | |
| - return err | |
| - } | |
| - if asOf.Less(min) { | |
| + ptp := tc.Server(i).ExecutorConfig().(sql.ExecutorConfig).ProtectedTimestampProvider | |
| + if ptp.Iterate(ctx, tableKey, tableKey, func(record *ptpb.Record) (wantMore bool) { | |
| + return false | |
| + }).Less(min) { | |
| return errors.Errorf("not yet read") | |
| } | |
| } | |
| @@ -3575,27 +3572,11 @@ func TestStrictGCEnforcement(t *testing.T) { | |
| } | |
| refreshPastLeaseStart = func(t *testing.T) { | |
| for i := 0; i < tc.NumServers(); i++ { | |
| - ptsReader := tc.GetFirstStoreFromServer(t, 0).GetStoreConfig().ProtectedTimestampReader | |
| - _, r := getFirstStoreReplica(t, tc.Server(i), tableKey) | |
| - l, _ := r.GetLease() | |
| - require.NoError( | |
| - t, | |
| - spanconfigptsreader.TestingRefreshPTSState(ctx, t, ptsReader, l.Start.ToTimestamp().Next()), | |
| - ) | |
| - err := r.ReadProtectedTimestamps(ctx) | |
| - require.NoError(t, err) | |
| - } | |
| - } | |
| - refreshCacheAndUpdatePTSState = func(t *testing.T, nodeID roachpb.NodeID) { | |
| - for i := 0; i < tc.NumServers(); i++ { | |
| - if tc.Server(i).NodeID() != nodeID { | |
| - continue | |
| - } | |
| ptp := tc.Server(i).ExecutorConfig().(sql.ExecutorConfig).ProtectedTimestampProvider | |
| - require.NoError(t, ptp.Refresh(ctx, tc.Server(i).Clock().Now())) | |
| _, r := getFirstStoreReplica(t, tc.Server(i), tableKey) | |
| - err := r.ReadProtectedTimestamps(ctx) | |
| - require.NoError(t, err) | |
| + l, _ := r.GetLease() | |
| + require.NoError(t, ptp.Refresh(ctx, l.Start.ToTimestamp().Next())) | |
| + r.ReadProtectedTimestamps(ctx) | |
| } | |
| } | |
| ) | |
| @@ -3651,15 +3632,13 @@ func TestStrictGCEnforcement(t *testing.T) { | |
| })) | |
| assertScanRejected(t) | |
| - desc, err := tc.LookupRange(tableKey) | |
| - require.NoError(t, err) | |
| - target, err := tc.FindRangeLeaseHolder(desc, nil) | |
| - require.NoError(t, err) | |
| - refreshCacheAndUpdatePTSState(t, target.NodeID) | |
| + require.NoError(t, ptp.Verify(ctx, rec.ID.GetUUID())) | |
| assertScanOk(t) | |
| // Transfer the lease and demonstrate that the query succeeds because we're | |
| // cautious in the face of lease transfers. | |
| + desc, err := tc.LookupRange(tableKey) | |
| + require.NoError(t, err) | |
| require.NoError(t, tc.TransferRangeLease(desc, tc.Target(1))) | |
| assertScanOk(t) | |
| }) | |
| diff --git a/pkg/kv/kvserver/client_spanconfigs_test.go b/pkg/kv/kvserver/client_spanconfigs_test.go | |
| index 3f46b364df..c9bfd59838 100644 | |
| --- a/pkg/kv/kvserver/client_spanconfigs_test.go | |
| +++ b/pkg/kv/kvserver/client_spanconfigs_test.go | |
| @@ -47,7 +47,6 @@ func TestSpanConfigUpdateAppliedToReplica(t *testing.T) { | |
| Store: &kvserver.StoreTestingKnobs{ | |
| DisableMergeQueue: true, | |
| DisableSplitQueue: true, | |
| - DisableGCQueue: true, | |
| }, | |
| SpanConfig: &spanconfig.TestingKnobs{ | |
| StoreKVSubscriberOverride: mockSubscriber, | |
| @@ -119,12 +118,6 @@ func (m *mockSpanConfigSubscriber) GetSpanConfigForKey( | |
| return m.Store.GetSpanConfigForKey(ctx, key) | |
| } | |
| -func (m *mockSpanConfigSubscriber) GetProtectionTimestamps( | |
| - context.Context, roachpb.Span, | |
| -) ([]hlc.Timestamp, hlc.Timestamp, error) { | |
| - panic("unimplemented") | |
| -} | |
| - | |
| func (m *mockSpanConfigSubscriber) LastUpdated() hlc.Timestamp { | |
| panic("unimplemented") | |
| } | |
| diff --git a/pkg/kv/kvserver/client_split_test.go b/pkg/kv/kvserver/client_split_test.go | |
| index 64430abd59..1847bf028a 100644 | |
| --- a/pkg/kv/kvserver/client_split_test.go | |
| +++ b/pkg/kv/kvserver/client_split_test.go | |
| @@ -38,7 +38,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/rpc" | |
| "github.com/cockroachdb/cockroach/pkg/server" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/bootstrap" | |
| "github.com/cockroachdb/cockroach/pkg/sql/catalog/systemschema" | |
| "github.com/cockroachdb/cockroach/pkg/storage" | |
| @@ -2385,9 +2384,6 @@ func TestUnsplittableRange(t *testing.T) { | |
| DefaultZoneConfigOverride: &zoneConfig, | |
| DefaultSystemZoneConfigOverride: &zoneSystemConfig, | |
| }, | |
| - SpanConfig: &spanconfig.TestingKnobs{ | |
| - ProtectedTSReaderOverrideFn: spanconfig.EmptyProtectedTSReader, | |
| - }, | |
| }, | |
| }) | |
| s := serv.(*server.TestServer) | |
| @@ -2433,7 +2429,9 @@ func TestUnsplittableRange(t *testing.T) { | |
| manualClock.Increment(10 * ttl.Nanoseconds()) | |
| // Trigger the MVCC GC queue, which should clean up the earlier version of the | |
| // row. Once the first version of the row is cleaned up, the range should | |
| - // exit the split queue purgatory. | |
| + // exit the split queue purgatory. We need to tickle the protected timestamp | |
| + // subsystem to release a timestamp at which we get to actually remove the data. | |
| + require.NoError(t, store.GetStoreConfig().ProtectedTimestampCache.Refresh(ctx, s.Clock().Now())) | |
| repl := store.LookupReplica(tableKey) | |
| if err := store.ManualMVCCGC(repl); err != nil { | |
| t.Fatal(err) | |
| diff --git a/pkg/kv/kvserver/closed_timestamp_test.go b/pkg/kv/kvserver/closed_timestamp_test.go | |
| index 862d12b248..55e7035853 100644 | |
| --- a/pkg/kv/kvserver/closed_timestamp_test.go | |
| +++ b/pkg/kv/kvserver/closed_timestamp_test.go | |
| @@ -606,8 +606,6 @@ func TestClosedTimestampFrozenAfterSubsumption(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| - skip.UnderRace(t) | |
| - | |
| for _, test := range []struct { | |
| name string | |
| // transferLease, if set, will be called while the RHS is subsumed in order | |
| diff --git a/pkg/kv/kvserver/concurrency/BUILD.bazel b/pkg/kv/kvserver/concurrency/BUILD.bazel | |
| index 76721b2f43..4e28adfbb8 100644 | |
| --- a/pkg/kv/kvserver/concurrency/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/concurrency/BUILD.bazel | |
| @@ -18,7 +18,6 @@ go_library( | |
| "//pkg/keys", | |
| "//pkg/kv", | |
| "//pkg/kv/kvserver/concurrency/lock", | |
| - "//pkg/kv/kvserver/concurrency/poison", | |
| "//pkg/kv/kvserver/intentresolver", | |
| "//pkg/kv/kvserver/spanlatch", | |
| "//pkg/kv/kvserver/spanset", | |
| @@ -52,11 +51,9 @@ go_test( | |
| ], | |
| data = glob(["testdata/**"]), | |
| embed = [":concurrency"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/kv/kvserver/batcheval", | |
| "//pkg/kv/kvserver/concurrency/lock", | |
| - "//pkg/kv/kvserver/concurrency/poison", | |
| "//pkg/kv/kvserver/intentresolver", | |
| "//pkg/kv/kvserver/spanlatch", | |
| "//pkg/kv/kvserver/spanset", | |
| diff --git a/pkg/kv/kvserver/concurrency/concurrency_control.go b/pkg/kv/kvserver/concurrency/concurrency_control.go | |
| index c5d27ebe1f..883a83e603 100644 | |
| --- a/pkg/kv/kvserver/concurrency/concurrency_control.go | |
| +++ b/pkg/kv/kvserver/concurrency/concurrency_control.go | |
| @@ -18,7 +18,6 @@ import ( | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock" | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/spanset" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/txnwait" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -190,15 +189,6 @@ type RequestSequencer interface { | |
| // does so, it will not return a request guard. | |
| SequenceReq(context.Context, *Guard, Request, RequestEvalKind) (*Guard, Response, *Error) | |
| - // PoisonReq idempotently marks a Guard as poisoned, indicating that its | |
| - // latches may be held for an indefinite amount of time. Requests waiting on | |
| - // this Guard will be notified. Latch acquisitions under poison.Policy_Error | |
| - // react to this by failing with a poison.PoisonedError, while requests under | |
| - // poison.Policy_Wait continue waiting, but propagate the poisoning upwards. | |
| - // | |
| - // See poison.Policy for details. | |
| - PoisonReq(*Guard) | |
| - | |
| // FinishReq marks the request as complete, releasing any protection | |
| // the request had against conflicting requests and allowing conflicting | |
| // requests that are blocked on this one to proceed. The guard should not | |
| @@ -395,9 +385,6 @@ type Request struct { | |
| // with a WriteIntentError instead of entering the queue and waiting. | |
| MaxLockWaitQueueLength int | |
| - // The poison.Policy to use for this Request. | |
| - PoisonPolicy poison.Policy | |
| - | |
| // The individual requests in the batch. | |
| Requests []roachpb.RequestUnion | |
| @@ -477,12 +464,9 @@ type latchManager interface { | |
| // WaitFor waits for conflicting latches on the specified spans without adding | |
| // any latches itself. Fast path for operations that only require flushing out | |
| // old operations without blocking any new ones. | |
| - WaitFor(ctx context.Context, spans *spanset.SpanSet, pp poison.Policy) *Error | |
| - | |
| - // Poison a guard's latches, allowing waiters to fail fast. | |
| - Poison(latchGuard) | |
| + WaitFor(ctx context.Context, spans *spanset.SpanSet) *Error | |
| - // Release a guard's latches, relinquish its protection from conflicting requests. | |
| + // Releases latches, relinquish its protection from conflicting requests. | |
| Release(latchGuard) | |
| // Metrics returns information about the state of the latchManager. | |
| diff --git a/pkg/kv/kvserver/concurrency/concurrency_manager.go b/pkg/kv/kvserver/concurrency/concurrency_manager.go | |
| index 5bc0d22414..c3ee4f1cd1 100644 | |
| --- a/pkg/kv/kvserver/concurrency/concurrency_manager.go | |
| +++ b/pkg/kv/kvserver/concurrency/concurrency_manager.go | |
| @@ -230,7 +230,7 @@ func (m *managerImpl) sequenceReqWithGuard(ctx context.Context, g *Guard) (Respo | |
| // them. | |
| if shouldWaitOnLatchesWithoutAcquiring(g.Req) { | |
| log.Event(ctx, "waiting on latches without acquiring") | |
| - return nil, m.lm.WaitFor(ctx, g.Req.LatchSpans, g.Req.PoisonPolicy) | |
| + return nil, m.lm.WaitFor(ctx, g.Req.LatchSpans) | |
| } | |
| // Provide the manager with an opportunity to intercept the request. It | |
| @@ -382,15 +382,6 @@ func shouldWaitOnLatchesWithoutAcquiring(req Request) bool { | |
| return req.isSingle(roachpb.Barrier) | |
| } | |
| -// PoisonReq implements the RequestSequencer interface. | |
| -func (m *managerImpl) PoisonReq(g *Guard) { | |
| - // NB: g.lg == nil is the case for requests that ignore latches, see | |
| - // shouldIgnoreLatches. | |
| - if g.lg != nil { | |
| - m.lm.Poison(g.lg) | |
| - } | |
| -} | |
| - | |
| // FinishReq implements the RequestSequencer interface. | |
| func (m *managerImpl) FinishReq(g *Guard) { | |
| // NOTE: we release latches _before_ exiting lock wait-queues deliberately. | |
| diff --git a/pkg/kv/kvserver/concurrency/concurrency_manager_test.go b/pkg/kv/kvserver/concurrency/concurrency_manager_test.go | |
| index cfe06d29e9..e533b51748 100644 | |
| --- a/pkg/kv/kvserver/concurrency/concurrency_manager_test.go | |
| +++ b/pkg/kv/kvserver/concurrency/concurrency_manager_test.go | |
| @@ -54,10 +54,9 @@ import ( | |
| // The input files use the following DSL: | |
| // | |
| // new-txn name=<txn-name> ts=<int>[,<int>] epoch=<int> [uncertainty-limit=<int>[,<int>]] | |
| -// new-request name=<req-name> txn=<txn-name>|none ts=<int>[,<int>] [priority] [inconsistent] [wait-policy=<policy>] [lock-timeout] [max-lock-wait-queue-length=<int>] [poison-policy=[err|wait]] | |
| +// new-request name=<req-name> txn=<txn-name>|none ts=<int>[,<int>] [priority] [inconsistent] [wait-policy=<policy>] [lock-timeout] [max-lock-wait-queue-length=<int>] | |
| // <proto-name> [<field-name>=<field-value>...] (hint: see scanSingleRequest) | |
| // sequence req=<req-name> [eval-kind=<pess|opt|pess-after-opt] | |
| -// poison req=<req-name> | |
| // finish req=<req-name> | |
| // | |
| // handle-write-intent-error req=<req-name> txn=<txn-name> key=<key> lease-seq=<seq> | |
| @@ -170,8 +169,6 @@ func TestConcurrencyManagerBasic(t *testing.T) { | |
| d.ScanArgs(t, "max-lock-wait-queue-length", &maxLockWaitQueueLength) | |
| } | |
| - pp := scanPoisonPolicy(t, d) | |
| - | |
| // Each roachpb.Request is provided on an indented line. | |
| reqs, reqUnions := scanRequests(t, d, c) | |
| latchSpans, lockSpans := c.collectSpans(t, txn, ts, reqs) | |
| @@ -187,7 +184,6 @@ func TestConcurrencyManagerBasic(t *testing.T) { | |
| Requests: reqUnions, | |
| LatchSpans: latchSpans, | |
| LockSpans: lockSpans, | |
| - PoisonPolicy: pp, | |
| } | |
| return "" | |
| @@ -261,20 +257,6 @@ func TestConcurrencyManagerBasic(t *testing.T) { | |
| }) | |
| return c.waitAndCollect(t, mon) | |
| - case "poison": | |
| - var reqName string | |
| - d.ScanArgs(t, "req", &reqName) | |
| - guard, ok := c.guardsByReqName[reqName] | |
| - if !ok { | |
| - d.Fatalf(t, "unknown request: %s", reqName) | |
| - } | |
| - | |
| - opName := fmt.Sprintf("poison %s", reqName) | |
| - mon.runSync(opName, func(ctx context.Context) { | |
| - log.Event(ctx, "poisoning request") | |
| - m.PoisonReq(guard) | |
| - }) | |
| - return c.waitAndCollect(t, mon) | |
| case "handle-write-intent-error": | |
| var reqName string | |
| d.ScanArgs(t, "req", &reqName) | |
| diff --git a/pkg/kv/kvserver/concurrency/datadriven_util_test.go b/pkg/kv/kvserver/concurrency/datadriven_util_test.go | |
| index 11067bce44..e9e8f9483a 100644 | |
| --- a/pkg/kv/kvserver/concurrency/datadriven_util_test.go | |
| +++ b/pkg/kv/kvserver/concurrency/datadriven_util_test.go | |
| @@ -15,7 +15,6 @@ import ( | |
| "testing" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock" | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/storage/enginepb" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| @@ -76,24 +75,6 @@ func scanWaitPolicy(t *testing.T, d *datadriven.TestData, required bool) lock.Wa | |
| } | |
| } | |
| -func scanPoisonPolicy(t *testing.T, d *datadriven.TestData) poison.Policy { | |
| - const key = "poison-policy" | |
| - if !d.HasArg(key) { | |
| - return poison.Policy_Error | |
| - } | |
| - var policy string | |
| - d.ScanArgs(t, key, &policy) | |
| - switch policy { | |
| - case "error": | |
| - return poison.Policy_Error | |
| - case "wait": | |
| - return poison.Policy_Wait | |
| - default: | |
| - d.Fatalf(t, "unknown poison policy: %s", policy) | |
| - return 0 | |
| - } | |
| -} | |
| - | |
| func scanSingleRequest( | |
| t *testing.T, d *datadriven.TestData, line string, txns map[string]*roachpb.Transaction, | |
| ) roachpb.Request { | |
| diff --git a/pkg/kv/kvserver/concurrency/latch_manager.go b/pkg/kv/kvserver/concurrency/latch_manager.go | |
| index b0f0705734..b0a4b8eb10 100644 | |
| --- a/pkg/kv/kvserver/concurrency/latch_manager.go | |
| +++ b/pkg/kv/kvserver/concurrency/latch_manager.go | |
| @@ -13,7 +13,6 @@ package concurrency | |
| import ( | |
| "context" | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/spanlatch" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/spanset" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -25,7 +24,7 @@ type latchManagerImpl struct { | |
| } | |
| func (m *latchManagerImpl) Acquire(ctx context.Context, req Request) (latchGuard, *Error) { | |
| - lg, err := m.m.Acquire(ctx, req.LatchSpans, req.PoisonPolicy) | |
| + lg, err := m.m.Acquire(ctx, req.LatchSpans) | |
| if err != nil { | |
| return nil, roachpb.NewError(err) | |
| } | |
| @@ -33,7 +32,7 @@ func (m *latchManagerImpl) Acquire(ctx context.Context, req Request) (latchGuard | |
| } | |
| func (m *latchManagerImpl) AcquireOptimistic(req Request) latchGuard { | |
| - lg := m.m.AcquireOptimistic(req.LatchSpans, req.PoisonPolicy) | |
| + lg := m.m.AcquireOptimistic(req.LatchSpans) | |
| return lg | |
| } | |
| @@ -51,20 +50,14 @@ func (m *latchManagerImpl) WaitUntilAcquired( | |
| return lg, nil | |
| } | |
| -func (m *latchManagerImpl) WaitFor( | |
| - ctx context.Context, ss *spanset.SpanSet, pp poison.Policy, | |
| -) *Error { | |
| - err := m.m.WaitFor(ctx, ss, pp) | |
| +func (m *latchManagerImpl) WaitFor(ctx context.Context, ss *spanset.SpanSet) *Error { | |
| + err := m.m.WaitFor(ctx, ss) | |
| if err != nil { | |
| return roachpb.NewError(err) | |
| } | |
| return nil | |
| } | |
| -func (m *latchManagerImpl) Poison(lg latchGuard) { | |
| - m.m.Poison(lg.(*spanlatch.Guard)) | |
| -} | |
| - | |
| func (m *latchManagerImpl) Release(lg latchGuard) { | |
| m.m.Release(lg.(*spanlatch.Guard)) | |
| } | |
| diff --git a/pkg/kv/kvserver/concurrency/lock/BUILD.bazel b/pkg/kv/kvserver/concurrency/lock/BUILD.bazel | |
| index 0c44ef2c00..dc7ca0cb25 100644 | |
| --- a/pkg/kv/kvserver/concurrency/lock/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/concurrency/lock/BUILD.bazel | |
| @@ -1,32 +1,21 @@ | |
| load("@rules_proto//proto:defs.bzl", "proto_library") | |
| load("@io_bazel_rules_go//proto:def.bzl", "go_proto_library") | |
| -load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test") | |
| +load("@io_bazel_rules_go//go:def.bzl", "go_library") | |
| go_library( | |
| name = "lock", | |
| - srcs = [ | |
| - "lock_waiter.go", | |
| - "locking.go", | |
| - ], | |
| + srcs = ["locking.go"], | |
| embed = [":lock_go_proto"], | |
| importpath = "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock", | |
| visibility = ["//visibility:public"], | |
| - deps = ["@com_github_cockroachdb_redact//:redact"], | |
| ) | |
| proto_library( | |
| name = "lock_proto", | |
| - srcs = [ | |
| - "lock_waiter.proto", | |
| - "locking.proto", | |
| - ], | |
| + srcs = ["locking.proto"], | |
| strip_import_prefix = "/pkg", | |
| visibility = ["//visibility:public"], | |
| - deps = [ | |
| - "//pkg/storage/enginepb:enginepb_proto", | |
| - "@com_github_gogo_protobuf//gogoproto:gogo_proto", | |
| - "@com_google_protobuf//:duration_proto", | |
| - ], | |
| + deps = ["@com_github_gogo_protobuf//gogoproto:gogo_proto"], | |
| ) | |
| go_proto_library( | |
| @@ -35,22 +24,5 @@ go_proto_library( | |
| importpath = "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock", | |
| proto = ":lock_proto", | |
| visibility = ["//visibility:public"], | |
| - deps = [ | |
| - "//pkg/storage/enginepb", | |
| - "@com_github_gogo_protobuf//gogoproto", | |
| - ], | |
| -) | |
| - | |
| -go_test( | |
| - name = "lock_test", | |
| - srcs = ["lock_waiter_test.go"], | |
| - deps = [ | |
| - ":lock", | |
| - "//pkg/roachpb", | |
| - "//pkg/storage/enginepb", | |
| - "//pkg/util/hlc", | |
| - "//pkg/util/uuid", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| - "@com_github_stretchr_testify//require", | |
| - ], | |
| + deps = ["@com_github_gogo_protobuf//gogoproto"], | |
| ) | |
| diff --git a/pkg/kv/kvserver/concurrency/lock/lock_waiter.go b/pkg/kv/kvserver/concurrency/lock/lock_waiter.go | |
| deleted file mode 100644 | |
| index be334625d4..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/lock/lock_waiter.go | |
| +++ /dev/null | |
| @@ -1,30 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -// Package lock provides type definitions for locking-related concepts used by | |
| -// concurrency control in the key-value layer. | |
| -package lock | |
| - | |
| -import "github.com/cockroachdb/redact" | |
| - | |
| -// SafeFormat implements redact.SafeFormatter. | |
| -func (lw Waiter) SafeFormat(w redact.SafePrinter, _ rune) { | |
| - expand := w.Flag('+') | |
| - | |
| - txnIDRedactableString := redact.Sprint(nil) | |
| - if lw.WaitingTxn != nil { | |
| - if expand { | |
| - txnIDRedactableString = redact.Sprint(lw.WaitingTxn.ID) | |
| - } else { | |
| - txnIDRedactableString = redact.Sprint(lw.WaitingTxn.Short()) | |
| - } | |
| - } | |
| - w.Printf("waiting_txn:%s active_waiter:%t strength:%s wait_duration:%s", txnIDRedactableString, lw.ActiveWaiter, lw.Strength, lw.WaitDuration) | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/lock/lock_waiter.proto b/pkg/kv/kvserver/concurrency/lock/lock_waiter.proto | |
| deleted file mode 100644 | |
| index e5bfce38ea..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/lock/lock_waiter.proto | |
| +++ /dev/null | |
| @@ -1,35 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -syntax = "proto3"; | |
| -package cockroach.kv.kvserver.concurrency.lock; | |
| -option go_package = "lock"; | |
| - | |
| -import "kv/kvserver/concurrency/lock/locking.proto"; | |
| -import "storage/enginepb/mvcc3.proto"; | |
| -import "gogoproto/gogo.proto"; | |
| -import "google/protobuf/duration.proto"; | |
| - | |
| -// Waiter represents a transaction (or non-transactional operation) that is | |
| -// waiting in the wait queue of readers or writers on an individual lock. | |
| -message Waiter { | |
| - // The transaction associated with this waiter, or nil in the case of a | |
| - // non-transactional waiter. | |
| - storage.enginepb.TxnMeta waiting_txn = 1; | |
| - // Represents if this operation is actively waiting on the lock. While all | |
| - // readers are active waiters, there are some cases in which writers may not | |
| - // be actively waiting, for instance in the case of a broken reservation. | |
| - bool active_waiter = 2; | |
| - // The strength at which this waiter is attempting to acquire the lock. | |
| - Strength strength = 3; | |
| - // The wall clock duration since this operation began waiting on the lock. | |
| - google.protobuf.Duration wait_duration = 4 [(gogoproto.nullable) = false, | |
| - (gogoproto.stdduration) = true]; | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/lock/lock_waiter_test.go b/pkg/kv/kvserver/concurrency/lock/lock_waiter_test.go | |
| deleted file mode 100644 | |
| index 32d01ce4e9..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/lock/lock_waiter_test.go | |
| +++ /dev/null | |
| @@ -1,79 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -// Package lock provides type definitions for locking-related concepts used by | |
| -// concurrency control in the key-value layer. | |
| -package lock_test | |
| - | |
| -import ( | |
| - "testing" | |
| - "time" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock" | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/storage/enginepb" | |
| - "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| - "github.com/cockroachdb/cockroach/pkg/util/uuid" | |
| - "github.com/cockroachdb/redact" | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -func TestWaiterSafeFormat(t *testing.T) { | |
| - ts := hlc.Timestamp{Logical: 1} | |
| - txnMeta := &enginepb.TxnMeta{ | |
| - Key: roachpb.Key("foo"), | |
| - ID: uuid.NamespaceDNS, | |
| - Epoch: 2, | |
| - WriteTimestamp: ts, | |
| - MinTimestamp: ts, | |
| - Priority: 957356782, | |
| - Sequence: 123, | |
| - CoordinatorNodeID: 3, | |
| - } | |
| - waiter := &lock.Waiter{ | |
| - WaitingTxn: txnMeta, | |
| - ActiveWaiter: true, | |
| - Strength: lock.Exclusive, | |
| - WaitDuration: 135 * time.Second, | |
| - } | |
| - | |
| - require.EqualValues(t, | |
| - "waiting_txn:6ba7b810 active_waiter:true strength:Exclusive wait_duration:2m15s", | |
| - redact.Sprint(waiter).StripMarkers()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:6ba7b810-9dad-11d1-80b4-00c04fd430c8 active_waiter:true strength:Exclusive wait_duration:2m15s", | |
| - redact.Sprintf("%+v", waiter).StripMarkers()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:6ba7b810 active_waiter:true strength:Exclusive wait_duration:2m15s", | |
| - redact.Sprint(waiter).Redact()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:‹×› active_waiter:true strength:Exclusive wait_duration:2m15s", | |
| - redact.Sprintf("%+v", waiter).Redact()) | |
| - | |
| - nonTxnWaiter := &lock.Waiter{ | |
| - WaitingTxn: nil, | |
| - ActiveWaiter: false, | |
| - Strength: lock.None, | |
| - WaitDuration: 17 * time.Millisecond, | |
| - } | |
| - | |
| - require.EqualValues(t, | |
| - "waiting_txn:<nil> active_waiter:false strength:None wait_duration:17ms", | |
| - redact.Sprint(nonTxnWaiter).StripMarkers()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:<nil> active_waiter:false strength:None wait_duration:17ms", | |
| - redact.Sprintf("%+v", nonTxnWaiter).StripMarkers()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:<nil> active_waiter:false strength:None wait_duration:17ms", | |
| - redact.Sprint(nonTxnWaiter).Redact()) | |
| - require.EqualValues(t, | |
| - "waiting_txn:<nil> active_waiter:false strength:None wait_duration:17ms", | |
| - redact.Sprintf("%+v", nonTxnWaiter).Redact()) | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/lock/locking.go b/pkg/kv/kvserver/concurrency/lock/locking.go | |
| index f22a010d50..75d665d8f0 100644 | |
| --- a/pkg/kv/kvserver/concurrency/lock/locking.go | |
| +++ b/pkg/kv/kvserver/concurrency/lock/locking.go | |
| @@ -25,11 +25,5 @@ func init() { | |
| } | |
| } | |
| -// SafeValue implements redact.SafeValue. | |
| -func (Strength) SafeValue() {} | |
| - | |
| -// SafeValue implements redact.SafeValue. | |
| -func (Durability) SafeValue() {} | |
| - | |
| // SafeValue implements redact.SafeValue. | |
| func (WaitPolicy) SafeValue() {} | |
| diff --git a/pkg/kv/kvserver/concurrency/lock_table_test.go b/pkg/kv/kvserver/concurrency/lock_table_test.go | |
| index b5a1e30ce1..c78ec55fb0 100644 | |
| --- a/pkg/kv/kvserver/concurrency/lock_table_test.go | |
| +++ b/pkg/kv/kvserver/concurrency/lock_table_test.go | |
| @@ -21,7 +21,6 @@ import ( | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/lock" | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/spanlatch" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/spanset" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| @@ -871,7 +870,7 @@ func doWork(ctx context.Context, item *workItem, e *workloadExecutor) error { | |
| // cancellation, the code makes sure to release latches when returning | |
| // early due to error. Otherwise other requests will get stuck and | |
| // group.Wait() will not return until the test times out. | |
| - lg, err = e.lm.Acquire(context.Background(), item.request.LatchSpans, poison.Policy_Error) | |
| + lg, err = e.lm.Acquire(context.Background(), item.request.LatchSpans) | |
| if err != nil { | |
| return err | |
| } | |
| @@ -1415,7 +1414,7 @@ func doBenchWork(item *benchWorkItem, env benchEnv, doneCh chan<- error) { | |
| var err error | |
| firstIter := true | |
| for { | |
| - if lg, err = env.lm.Acquire(context.Background(), item.LatchSpans, poison.Policy_Error); err != nil { | |
| + if lg, err = env.lm.Acquire(context.Background(), item.LatchSpans); err != nil { | |
| doneCh <- err | |
| return | |
| } | |
| @@ -1450,7 +1449,7 @@ func doBenchWork(item *benchWorkItem, env benchEnv, doneCh chan<- error) { | |
| return | |
| } | |
| // Release locks. | |
| - if lg, err = env.lm.Acquire(context.Background(), item.LatchSpans, poison.Policy_Error); err != nil { | |
| + if lg, err = env.lm.Acquire(context.Background(), item.LatchSpans); err != nil { | |
| doneCh <- err | |
| return | |
| } | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/BUILD.bazel b/pkg/kv/kvserver/concurrency/poison/BUILD.bazel | |
| deleted file mode 100644 | |
| index cf282b9e15..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/BUILD.bazel | |
| +++ /dev/null | |
| @@ -1,61 +0,0 @@ | |
| -load("@rules_proto//proto:defs.bzl", "proto_library") | |
| -load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test") | |
| -load("@io_bazel_rules_go//proto:def.bzl", "go_proto_library") | |
| - | |
| -proto_library( | |
| - name = "poison_proto", | |
| - srcs = [ | |
| - "error.proto", | |
| - "policy.proto", | |
| - ], | |
| - strip_import_prefix = "/pkg", | |
| - visibility = ["//visibility:public"], | |
| - deps = [ | |
| - "//pkg/roachpb:roachpb_proto", | |
| - "//pkg/util/hlc:hlc_proto", | |
| - "@com_github_gogo_protobuf//gogoproto:gogo_proto", | |
| - ], | |
| -) | |
| - | |
| -go_proto_library( | |
| - name = "poison_go_proto", | |
| - compilers = ["//pkg/cmd/protoc-gen-gogoroach:protoc-gen-gogoroach_compiler"], | |
| - importpath = "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison", | |
| - proto = ":poison_proto", | |
| - visibility = ["//visibility:public"], | |
| - deps = [ | |
| - "//pkg/roachpb", | |
| - "//pkg/util/hlc", | |
| - "@com_github_gogo_protobuf//gogoproto", | |
| - ], | |
| -) | |
| - | |
| -go_library( | |
| - name = "poison", | |
| - srcs = ["error.go"], | |
| - embed = [":poison_go_proto"], | |
| - importpath = "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison", | |
| - visibility = ["//visibility:public"], | |
| - deps = [ | |
| - "//pkg/roachpb", | |
| - "//pkg/util/hlc", | |
| - "@com_github_cockroachdb_errors//:errors", | |
| - ], | |
| -) | |
| - | |
| -go_test( | |
| - name = "poison_test", | |
| - srcs = ["error_test.go"], | |
| - data = glob(["testdata/**"]), | |
| - deps = [ | |
| - ":poison", | |
| - "//pkg/keys", | |
| - "//pkg/roachpb", | |
| - "//pkg/testutils/echotest", | |
| - "//pkg/util/hlc", | |
| - "//pkg/util/leaktest", | |
| - "@com_github_cockroachdb_errors//:errors", | |
| - "@com_github_cockroachdb_redact//:redact", | |
| - "@com_github_stretchr_testify//require", | |
| - ], | |
| -) | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/error.go b/pkg/kv/kvserver/concurrency/poison/error.go | |
| deleted file mode 100644 | |
| index 3c473fc40a..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/error.go | |
| +++ /dev/null | |
| @@ -1,42 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package poison | |
| - | |
| -import ( | |
| - "fmt" | |
| - | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| - "github.com/cockroachdb/errors" | |
| -) | |
| - | |
| -// NewPoisonedError instantiates a *PoisonedError referencing a poisoned latch | |
| -// (as identified by span and timestamp). | |
| -func NewPoisonedError(span roachpb.Span, ts hlc.Timestamp) *PoisonedError { | |
| - return &PoisonedError{Span: span, Timestamp: ts} | |
| -} | |
| - | |
| -var _ errors.SafeFormatter = (*PoisonedError)(nil) | |
| -var _ fmt.Formatter = (*PoisonedError)(nil) | |
| - | |
| -// SafeFormatError implements errors.SafeFormatter. | |
| -func (e *PoisonedError) SafeFormatError(p errors.Printer) error { | |
| - p.Printf("encountered poisoned latch %s@%s", e.Span, e.Timestamp) | |
| - return nil | |
| -} | |
| - | |
| -// Format implements fmt.Formatter. | |
| -func (e *PoisonedError) Format(s fmt.State, verb rune) { errors.FormatError(e, s, verb) } | |
| - | |
| -// Error implements error. | |
| -func (e *PoisonedError) Error() string { | |
| - return fmt.Sprint(e) | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/error.proto b/pkg/kv/kvserver/concurrency/poison/error.proto | |
| deleted file mode 100644 | |
| index 3ebe47586f..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/error.proto | |
| +++ /dev/null | |
| @@ -1,26 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -syntax = "proto3"; | |
| -package cockroach.kv.kvserver.concurrency.poison; | |
| -option go_package = "poison"; | |
| - | |
| -import "util/hlc/timestamp.proto"; | |
| -import "roachpb/data.proto"; | |
| -import "gogoproto/gogo.proto"; | |
| - | |
| -// PoisonedError indicates that a request failed fast during sequencing as a | |
| -// result of having encountered a poisoned latch under Policy_Error. | |
| -// | |
| -// See also concurrency.RequestSequencer. | |
| -message PoisonedError { | |
| - roachpb.Span span = 1 [(gogoproto.nullable) = false]; | |
| - util.hlc.Timestamp timestamp = 2 [(gogoproto.nullable) = false]; | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/error_test.go b/pkg/kv/kvserver/concurrency/poison/error_test.go | |
| deleted file mode 100644 | |
| index 360501d4e6..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/error_test.go | |
| +++ /dev/null | |
| @@ -1,39 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -package poison_test | |
| - | |
| -import ( | |
| - "context" | |
| - "path/filepath" | |
| - "testing" | |
| - | |
| - _ "github.com/cockroachdb/cockroach/pkg/keys" // to init roachpb.PrettyPrintRange | |
| - "github.com/cockroachdb/cockroach/pkg/kv/kvserver/concurrency/poison" | |
| - "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/testutils/echotest" | |
| - "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| - "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| - "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| - "github.com/stretchr/testify/require" | |
| -) | |
| - | |
| -func TestPoisonedError(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - ctx := context.Background() | |
| - err := errors.DecodeError(ctx, errors.EncodeError(ctx, poison.NewPoisonedError( | |
| - roachpb.Span{Key: roachpb.Key("a")}, hlc.Timestamp{WallTime: 1}, | |
| - ))) | |
| - require.True(t, errors.HasType(err, (*poison.PoisonedError)(nil)), "%+v", err) | |
| - var buf redact.StringBuilder | |
| - buf.Printf("%s", err) | |
| - echotest.Require(t, string(buf.RedactableString()), filepath.Join("testdata", "poisoned_error.txt")) | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/policy.proto b/pkg/kv/kvserver/concurrency/poison/policy.proto | |
| deleted file mode 100644 | |
| index 5f3371cbf2..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/policy.proto | |
| +++ /dev/null | |
| @@ -1,35 +0,0 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| -// | |
| -// Use of this software is governed by the Business Source License | |
| -// included in the file licenses/BSL.txt. | |
| -// | |
| -// As of the Change Date specified in that file, in accordance with | |
| -// the Business Source License, use of this software will be governed | |
| -// by the Apache License, Version 2.0, included in the file | |
| -// licenses/APL.txt. | |
| - | |
| -syntax = "proto3"; | |
| -package cockroach.kv.kvserver.concurrency.poison; | |
| -option go_package = "poison"; | |
| - | |
| -import "gogoproto/gogo.proto"; | |
| - | |
| -// Policy determines how a request will react to encountering a poisoned | |
| -// latch. A poisoned latch is a latch for which the holder is unable to make | |
| -// progress. That is, waiters of this latch should not expect to be able to | |
| -// acquire this latch "for some time"; in practice this is the case of an | |
| -// unavailable Replica. | |
| -// | |
| -// The name is inspired by Rust's mutexes, which undergo poisoning[^1] when a | |
| -// thread panics while holding the mutex. | |
| -// | |
| -// [^1]: https://doc.rust-lang.org/std/sync/struct.Mutex.html#poisoning | |
| -enum Policy { | |
| - // Policy_Wait instructs a request to return an error upon encountering | |
| - // a poisoned latch. | |
| - Wait = 0; | |
| - | |
| - // Policy_Error instructs a request to return an error upon encountering | |
| - // a poisoned latch. | |
| - Error = 1; | |
| -} | |
| diff --git a/pkg/kv/kvserver/concurrency/poison/testdata/poisoned_error.txt b/pkg/kv/kvserver/concurrency/poison/testdata/poisoned_error.txt | |
| deleted file mode 100644 | |
| index b4b6bf4061..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/poison/testdata/poisoned_error.txt | |
| +++ /dev/null | |
| @@ -1,3 +0,0 @@ | |
| -echo | |
| ----- | |
| -encountered poisoned latch ‹a›@0.000000001,0 | |
| diff --git a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err b/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err | |
| deleted file mode 100644 | |
| index 2fffae01a7..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err | |
| +++ /dev/null | |
| @@ -1,60 +0,0 @@ | |
| -# This test sets up the following situation: | |
| -# | |
| -# e <- put (PoisonPolicyErr; waiting) | |
| -# b---f <- scan (PoisonPolicyErr; waiting) | |
| -# c <- put (PoisonPolicyErr; sequenced, poisoned) | |
| -# | |
| -# Since everyone uses PoisonPolicyErr, the chain unwinds. However, only `b---f` | |
| -# gets an error, since it overlaps `c`. `e` can proceed once `c` and `b---f` | |
| -# have finished. | |
| - | |
| -new-request txn=none name=putc ts=10,0 | |
| - put key=c value=hi | |
| ----- | |
| - | |
| -sequence req=putc | |
| ----- | |
| -[1] sequence putc: sequencing request | |
| -[1] sequence putc: acquiring latches | |
| -[1] sequence putc: scanning lock table for conflicting locks | |
| -[1] sequence putc: sequencing complete, returned guard | |
| - | |
| -new-request txn=none name=readbf ts=11,1 | |
| - scan key=b endkey=f | |
| ----- | |
| - | |
| -sequence req=readbf | |
| ----- | |
| -[2] sequence readbf: sequencing request | |
| -[2] sequence readbf: acquiring latches | |
| -[2] sequence readbf: waiting to acquire read latch {b-f}@11.000000000,1, held by write latch [email protected],0 | |
| -[2] sequence readbf: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -new-request txn=none name=pute ts=11,0 | |
| - put key=e value=hi | |
| ----- | |
| - | |
| -sequence req=pute | |
| ----- | |
| -[3] sequence pute: sequencing request | |
| -[3] sequence pute: acquiring latches | |
| -[3] sequence pute: waiting to acquire write latch [email protected],0, held by read latch {b-f}@11.000000000,1 | |
| -[3] sequence pute: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -poison req=putc | |
| ----- | |
| -[-] poison putc: poisoning request | |
| -[2] sequence readbf: sequencing complete, returned error: encountered poisoned latch [email protected],0 | |
| -[3] sequence pute: scanning lock table for conflicting locks | |
| -[3] sequence pute: sequencing complete, returned guard | |
| - | |
| -finish req=putc | |
| ----- | |
| -[-] finish putc: finishing request | |
| - | |
| -finish req=pute | |
| ----- | |
| -[-] finish pute: finishing request | |
| - | |
| -reset | |
| ----- | |
| diff --git a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err_indirect b/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err_indirect | |
| deleted file mode 100644 | |
| index 2615d3524c..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_err_indirect | |
| +++ /dev/null | |
| @@ -1,61 +0,0 @@ | |
| -# This test sets up the following situation: | |
| -# | |
| -# e <- put (PoisonPolicyError; waiting) | |
| -# b---f <- scan (PoisonPolicyWait; waiting) | |
| -# c <- put (PoisonPolicyWait; sequenced, poisoned) | |
| -# | |
| -# When `c` gets poisoned (and continues waiting), the same | |
| -# happens to `b---f`, which prompts `e` to fail fast. | |
| - | |
| -new-request txn=none name=putc ts=10,0 poison-policy=wait | |
| - put key=c value=hi | |
| ----- | |
| - | |
| -sequence req=putc | |
| ----- | |
| -[1] sequence putc: sequencing request | |
| -[1] sequence putc: acquiring latches | |
| -[1] sequence putc: scanning lock table for conflicting locks | |
| -[1] sequence putc: sequencing complete, returned guard | |
| - | |
| -new-request txn=none name=readbf ts=11,1 poison-policy=wait | |
| - scan key=b endkey=f | |
| ----- | |
| - | |
| -sequence req=readbf | |
| ----- | |
| -[2] sequence readbf: sequencing request | |
| -[2] sequence readbf: acquiring latches | |
| -[2] sequence readbf: waiting to acquire read latch {b-f}@11.000000000,1, held by write latch [email protected],0 | |
| -[2] sequence readbf: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -new-request txn=none name=pute ts=11,0 | |
| - put key=e value=hi | |
| ----- | |
| - | |
| -sequence req=pute | |
| ----- | |
| -[3] sequence pute: sequencing request | |
| -[3] sequence pute: acquiring latches | |
| -[3] sequence pute: waiting to acquire write latch [email protected],0, held by read latch {b-f}@11.000000000,1 | |
| -[3] sequence pute: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -poison req=putc | |
| ----- | |
| -[-] poison putc: poisoning request | |
| -[2] sequence readbf: encountered poisoned latch; continuing to wait | |
| -[2] sequence readbf: blocked on select in spanlatch.(*Manager).waitForSignal | |
| -[3] sequence pute: sequencing complete, returned error: encountered poisoned latch {b-f}@11.000000000,1 | |
| - | |
| -finish req=putc | |
| ----- | |
| -[-] finish putc: finishing request | |
| -[2] sequence readbf: scanning lock table for conflicting locks | |
| -[2] sequence readbf: sequencing complete, returned guard | |
| - | |
| -finish req=readbf | |
| ----- | |
| -[-] finish readbf: finishing request | |
| - | |
| -reset | |
| ----- | |
| diff --git a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_disjoint b/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_disjoint | |
| deleted file mode 100644 | |
| index bf7646ff84..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_disjoint | |
| +++ /dev/null | |
| @@ -1,59 +0,0 @@ | |
| -# This test sets up the following situation: | |
| -# | |
| -# e <- put (PoisonPolicyWait; waiting) | |
| -# b---f <- scan (PoisonPolicyError; waiting) | |
| -# c <- put (PoisonPolicyWait; sequenced, poisoned) | |
| -# | |
| -# The top and bottom request use PoisonPolicyWait, so the scan returns when `c` | |
| -# is poisoned, which in turn lets `e` through. However, `c` continues to wait. | |
| - | |
| -new-request txn=none name=putc ts=10,0 poison-policy=wait | |
| - put key=c value=hi | |
| ----- | |
| - | |
| -sequence req=putc | |
| ----- | |
| -[1] sequence putc: sequencing request | |
| -[1] sequence putc: acquiring latches | |
| -[1] sequence putc: scanning lock table for conflicting locks | |
| -[1] sequence putc: sequencing complete, returned guard | |
| - | |
| -new-request txn=none name=readbf ts=11,1 | |
| - scan key=b endkey=f | |
| ----- | |
| - | |
| -sequence req=readbf | |
| ----- | |
| -[2] sequence readbf: sequencing request | |
| -[2] sequence readbf: acquiring latches | |
| -[2] sequence readbf: waiting to acquire read latch {b-f}@11.000000000,1, held by write latch [email protected],0 | |
| -[2] sequence readbf: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -new-request txn=none name=pute ts=11,0 poison-policy=wait | |
| - put key=e value=hi | |
| ----- | |
| - | |
| -sequence req=pute | |
| ----- | |
| -[3] sequence pute: sequencing request | |
| -[3] sequence pute: acquiring latches | |
| -[3] sequence pute: waiting to acquire write latch [email protected],0, held by read latch {b-f}@11.000000000,1 | |
| -[3] sequence pute: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -poison req=putc | |
| ----- | |
| -[-] poison putc: poisoning request | |
| -[2] sequence readbf: sequencing complete, returned error: encountered poisoned latch [email protected],0 | |
| -[3] sequence pute: scanning lock table for conflicting locks | |
| -[3] sequence pute: sequencing complete, returned guard | |
| - | |
| -finish req=putc | |
| ----- | |
| -[-] finish putc: finishing request | |
| - | |
| -finish req=pute | |
| ----- | |
| -[-] finish pute: finishing request | |
| - | |
| -reset | |
| ----- | |
| diff --git a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_overlapping b/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_overlapping | |
| deleted file mode 100644 | |
| index 4e8799dde3..0000000000 | |
| --- a/pkg/kv/kvserver/concurrency/testdata/concurrency_manager/poison_policy_wait_overlapping | |
| +++ /dev/null | |
| @@ -1,61 +0,0 @@ | |
| -# This test sets up the following situation: | |
| -# | |
| -# c <- put (PoisonPolicyWait; waiting) | |
| -# b---f <- scan (PoisonPolicyError; waiting) | |
| -# c <- put (PoisonPolicyWait; sequenced, poisoned) | |
| -# | |
| -# When the bottom `c` is poisoned, `b---f` fails fast, and | |
| -# the top `c` poisons itself but continues to wait. | |
| - | |
| -new-request txn=none name=put1 ts=10,0 poison-policy=wait | |
| - put key=c value=hi | |
| ----- | |
| - | |
| -sequence req=put1 | |
| ----- | |
| -[1] sequence put1: sequencing request | |
| -[1] sequence put1: acquiring latches | |
| -[1] sequence put1: scanning lock table for conflicting locks | |
| -[1] sequence put1: sequencing complete, returned guard | |
| - | |
| -new-request txn=none name=readbf ts=11,1 | |
| - scan key=b endkey=f | |
| ----- | |
| - | |
| -sequence req=readbf | |
| ----- | |
| -[2] sequence readbf: sequencing request | |
| -[2] sequence readbf: acquiring latches | |
| -[2] sequence readbf: waiting to acquire read latch {b-f}@11.000000000,1, held by write latch [email protected],0 | |
| -[2] sequence readbf: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -new-request txn=none name=put2 ts=11,0 poison-policy=wait | |
| - put key=c value=bar | |
| ----- | |
| - | |
| -sequence req=put2 | |
| ----- | |
| -[3] sequence put2: sequencing request | |
| -[3] sequence put2: acquiring latches | |
| -[3] sequence put2: waiting to acquire write latch [email protected],0, held by write latch [email protected],0 | |
| -[3] sequence put2: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -poison req=put1 | |
| ----- | |
| -[-] poison put1: poisoning request | |
| -[2] sequence readbf: sequencing complete, returned error: encountered poisoned latch [email protected],0 | |
| -[3] sequence put2: encountered poisoned latch; continuing to wait | |
| -[3] sequence put2: blocked on select in spanlatch.(*Manager).waitForSignal | |
| - | |
| -finish req=put1 | |
| ----- | |
| -[-] finish put1: finishing request | |
| -[3] sequence put2: scanning lock table for conflicting locks | |
| -[3] sequence put2: sequencing complete, returned guard | |
| - | |
| -finish req=put2 | |
| ----- | |
| -[-] finish put2: finishing request | |
| - | |
| -reset | |
| ----- | |
| diff --git a/pkg/kv/kvserver/helpers_test.go b/pkg/kv/kvserver/helpers_test.go | |
| index 2d0156d689..a6e9218302 100644 | |
| --- a/pkg/kv/kvserver/helpers_test.go | |
| +++ b/pkg/kv/kvserver/helpers_test.go | |
| @@ -227,6 +227,13 @@ func (r *Replica) Breaker() *circuit2.Breaker { | |
| return r.breaker.wrapped | |
| } | |
| +func (r *Replica) VisitBreakerContexts(fn func(ctx context.Context)) { | |
| + r.breaker.cancels.Visit(func(ctx context.Context, _ func()) (remove bool) { | |
| + fn(ctx) | |
| + return false // keep | |
| + }) | |
| +} | |
| + | |
| func (r *Replica) AssertState(ctx context.Context, reader storage.Reader) { | |
| r.raftMu.Lock() | |
| defer r.raftMu.Unlock() | |
| @@ -476,14 +483,12 @@ func (r *Replica) MaybeUnquiesceAndWakeLeader() bool { | |
| return r.maybeUnquiesceAndWakeLeaderLocked() | |
| } | |
| -func (r *Replica) ReadProtectedTimestamps(ctx context.Context) error { | |
| +func (r *Replica) ReadProtectedTimestamps(ctx context.Context) { | |
| var ts cachedProtectedTimestampState | |
| defer r.maybeUpdateCachedProtectedTS(&ts) | |
| r.mu.RLock() | |
| defer r.mu.RUnlock() | |
| - var err error | |
| - ts, err = r.readProtectedTimestampsRLocked(ctx) | |
| - return err | |
| + ts = r.readProtectedTimestampsRLocked(ctx, nil /* f */) | |
| } | |
| // ClosedTimestampPolicy returns the closed timestamp policy of the range, which | |
| @@ -494,11 +499,6 @@ func (r *Replica) ClosedTimestampPolicy() roachpb.RangeClosedTimestampPolicy { | |
| return r.closedTimestampPolicyRLocked() | |
| } | |
| -// TripBreaker synchronously trips the breaker. | |
| -func (r *Replica) TripBreaker() { | |
| - r.breaker.tripSync(errors.New("injected error")) | |
| -} | |
| - | |
| // GetCircuitBreaker returns the circuit breaker controlling | |
| // connection attempts to the specified node. | |
| func (t *RaftTransport) GetCircuitBreaker( | |
| diff --git a/pkg/kv/kvserver/kvserverbase/bulk_adder.go b/pkg/kv/kvserver/kvserverbase/bulk_adder.go | |
| index c1a1c4470d..fc1b1f2798 100644 | |
| --- a/pkg/kv/kvserver/kvserverbase/bulk_adder.go | |
| +++ b/pkg/kv/kvserver/kvserverbase/bulk_adder.go | |
| @@ -25,6 +25,16 @@ type BulkAdderOptions struct { | |
| // behalf of which it is adding data. | |
| Name string | |
| + // SSTSize is the size at which an SST will be flushed and a new one started. | |
| + // SSTs are also split during a buffer flush to avoid spanning range bounds so | |
| + // they may be smaller than this limit. | |
| + SSTSize func() int64 | |
| + | |
| + // SplitAndScatterAfter is the number of bytes which if added without hitting | |
| + // an existing split will cause the adder to split and scatter the next span. | |
| + // A function returning -1 is interpreted as indicating not to split. | |
| + SplitAndScatterAfter func() int64 | |
| + | |
| // MinBufferSize is the initial size of the BulkAdder buffer. It indicates the | |
| // amount of memory we require to be able to buffer data before flushing for | |
| // SST creation. | |
| @@ -74,6 +84,10 @@ type BulkAdderOptions struct { | |
| InitialSplitsIfUnordered int | |
| } | |
| +// DisableExplicitSplits can be returned by a SplitAndScatterAfter function to | |
| +// indicate that the SSTBatcher should not issue explicit splits. | |
| +const DisableExplicitSplits = -1 | |
| + | |
| // BulkAdderFactory describes a factory function for BulkAdders. | |
| type BulkAdderFactory func( | |
| ctx context.Context, db *kv.DB, timestamp hlc.Timestamp, opts BulkAdderOptions, | |
| diff --git a/pkg/kv/kvserver/loqrecovery/loqrecoverypb/recovery.proto b/pkg/kv/kvserver/loqrecovery/loqrecoverypb/recovery.proto | |
| index 0ec5262077..fe0427288a 100644 | |
| --- a/pkg/kv/kvserver/loqrecovery/loqrecoverypb/recovery.proto | |
| +++ b/pkg/kv/kvserver/loqrecovery/loqrecoverypb/recovery.proto | |
| @@ -47,7 +47,7 @@ message ReplicaInfo { | |
| uint64 raft_applied_index = 4; | |
| uint64 raft_committed_index = 5; | |
| repeated DescriptorChangeInfo raft_log_descriptor_changes = 6 [(gogoproto.nullable) = false, | |
| - (gogoproto.jsontag) = "raft_log_descriptor_changes,omitempty"]; | |
| + (gogoproto.jsontag) = ",omitempty"]; | |
| } | |
| // Collection of replica information gathered from a collect-info run on a single node. | |
| diff --git a/pkg/kv/kvserver/main_test.go b/pkg/kv/kvserver/main_test.go | |
| index 81e3d2bd7a..cb2ceefbb1 100644 | |
| --- a/pkg/kv/kvserver/main_test.go | |
| +++ b/pkg/kv/kvserver/main_test.go | |
| @@ -87,8 +87,7 @@ func TestMain(m *testing.M) { | |
| delete(notBelowRaftProtos, reflect.TypeOf(&roachpb.InternalTimeSeriesData{})) | |
| delete(notBelowRaftProtos, reflect.TypeOf(&enginepb.MVCCMetadataSubsetForMergeSerialization{})) | |
| for typ := range notBelowRaftProtos { | |
| - // NB: don't set failed=true. In a bazel world, we may just end up sharding in a way that | |
| - // doesn't observe some of the protos below raft. | |
| + failed = true | |
| fmt.Printf("%s: not observed below raft!\n", typ) | |
| } | |
| diff --git a/pkg/kv/kvserver/mvcc_gc_queue.go b/pkg/kv/kvserver/mvcc_gc_queue.go | |
| index d218a1389a..de016745c2 100644 | |
| --- a/pkg/kv/kvserver/mvcc_gc_queue.go | |
| +++ b/pkg/kv/kvserver/mvcc_gc_queue.go | |
| @@ -180,11 +180,7 @@ func (mgcq *mvccGCQueue) shouldQueue( | |
| // Consult the protected timestamp state to determine whether we can GC and | |
| // the timestamp which can be used to calculate the score. | |
| _, conf := repl.DescAndSpanConfig() | |
| - canGC, _, gcTimestamp, oldThreshold, newThreshold, err := repl.checkProtectedTimestampsForGC(ctx, conf.TTL()) | |
| - if err != nil { | |
| - log.VErrEventf(ctx, 2, "failed to check protected timestamp for gc: %v", err) | |
| - return false, 0 | |
| - } | |
| + canGC, _, gcTimestamp, oldThreshold, newThreshold := repl.checkProtectedTimestampsForGC(ctx, conf.TTL()) | |
| if !canGC { | |
| return false, 0 | |
| } | |
| @@ -529,10 +525,7 @@ func (mgcq *mvccGCQueue) process( | |
| // Consult the protected timestamp state to determine whether we can GC and | |
| // the timestamp which can be used to calculate the score and updated GC | |
| // threshold. | |
| - canGC, cacheTimestamp, gcTimestamp, oldThreshold, newThreshold, err := repl.checkProtectedTimestampsForGC(ctx, conf.TTL()) | |
| - if err != nil { | |
| - return false, err | |
| - } | |
| + canGC, cacheTimestamp, gcTimestamp, oldThreshold, newThreshold := repl.checkProtectedTimestampsForGC(ctx, conf.TTL()) | |
| if !canGC { | |
| return false, nil | |
| } | |
| diff --git a/pkg/kv/kvserver/node_liveness_test.go b/pkg/kv/kvserver/node_liveness_test.go | |
| index f4c04407b3..8e96674258 100644 | |
| --- a/pkg/kv/kvserver/node_liveness_test.go | |
| +++ b/pkg/kv/kvserver/node_liveness_test.go | |
| @@ -1020,7 +1020,7 @@ func TestNodeLivenessRetryAmbiguousResultError(t *testing.T) { | |
| if val := injectError.Load(); val != nil && val.(bool) { | |
| atomic.AddInt32(&injectedErrorCount, 1) | |
| injectError.Store(false) | |
| - return roachpb.NewError(roachpb.NewAmbiguousResultErrorf("test")) | |
| + return roachpb.NewError(roachpb.NewAmbiguousResultError("test")) | |
| } | |
| return nil | |
| } | |
| @@ -1063,7 +1063,7 @@ func TestNodeLivenessRetryAmbiguousResultOnCreateError(t *testing.T) { | |
| defer log.Scope(t).Close(t) | |
| errorsToTest := []error{ | |
| - roachpb.NewAmbiguousResultErrorf("test"), | |
| + roachpb.NewAmbiguousResultError("test"), | |
| roachpb.NewTransactionStatusError(roachpb.TransactionStatusError_REASON_UNKNOWN, "foo"), | |
| kv.OnePCNotAllowedError{}, | |
| } | |
| @@ -1207,7 +1207,8 @@ func TestNodeLivenessNoRetryOnAmbiguousResultCausedByCancellation(t *testing.T) | |
| // Check that Heartbeat() returned an ambiguous error, and take that as proof | |
| // that the heartbeat wasn't retried. | |
| - require.True(t, errors.HasType(err, (*roachpb.AmbiguousResultError)(nil)), "%+v", err) | |
| + require.Error(t, err) | |
| + require.Equal(t, "result is ambiguous (context done during DistSender.Send: context canceled)", err.Error()) | |
| } | |
| func verifyNodeIsDecommissioning(t *testing.T, tc *testcluster.TestCluster, nodeID roachpb.NodeID) { | |
| diff --git a/pkg/kv/kvserver/protectedts/BUILD.bazel b/pkg/kv/kvserver/protectedts/BUILD.bazel | |
| index 85d31485cc..e0b055fb23 100644 | |
| --- a/pkg/kv/kvserver/protectedts/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/protectedts/BUILD.bazel | |
| @@ -15,7 +15,6 @@ go_library( | |
| "//pkg/kv/kvserver/protectedts/ptpb", | |
| "//pkg/roachpb", | |
| "//pkg/settings", | |
| - "//pkg/spanconfig", | |
| "//pkg/util/hlc", | |
| "//pkg/util/metric", | |
| "//pkg/util/stop", | |
| diff --git a/pkg/kv/kvserver/protectedts/protectedts.go b/pkg/kv/kvserver/protectedts/protectedts.go | |
| index 2fedb889f3..10349f7e9a 100644 | |
| --- a/pkg/kv/kvserver/protectedts/protectedts.go | |
| +++ b/pkg/kv/kvserver/protectedts/protectedts.go | |
| @@ -18,7 +18,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/kv" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptpb" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| - "github.com/cockroachdb/cockroach/pkg/spanconfig" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/cockroach/pkg/util/metric" | |
| "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| @@ -39,6 +38,7 @@ var ErrExists = errors.New("protected timestamp record already exists") | |
| type Provider interface { | |
| Storage | |
| Cache | |
| + Verifier | |
| Reconciler | |
| Start(context.Context, *stop.Stopper) error | |
| @@ -115,7 +115,6 @@ type Iterator func(*ptpb.Record) (wantMore bool) | |
| // by any Records at a given asOf can move its GC threshold up to that | |
| // timestamp less its GC TTL. | |
| type Cache interface { | |
| - spanconfig.ProtectedTSReader | |
| // Iterate examines the records with spans which overlap with [from, to). | |
| // Nil values for from or to are equivalent to Key{}. The order of records | |
| @@ -133,6 +132,16 @@ type Cache interface { | |
| Refresh(_ context.Context, asOf hlc.Timestamp) error | |
| } | |
| +// Verifier provides a mechanism to verify that a created Record will certainly | |
| +// apply. | |
| +type Verifier interface { | |
| + | |
| + // Verify returns an error if the record of the provided ID cannot be | |
| + // verified. If nil is returned then the record has been proven to apply | |
| + // until it is removed. | |
| + Verify(context.Context, uuid.UUID) error | |
| +} | |
| + | |
| // Reconciler provides a mechanism to reconcile protected timestamp records with | |
| // external state. | |
| type Reconciler interface { | |
| @@ -169,9 +178,3 @@ func (c *emptyCache) QueryRecord( | |
| func (c *emptyCache) Refresh(_ context.Context, asOf hlc.Timestamp) error { | |
| return nil | |
| } | |
| - | |
| -func (c *emptyCache) GetProtectionTimestamps( | |
| - context.Context, roachpb.Span, | |
| -) (protectionTimestamps []hlc.Timestamp, asOf hlc.Timestamp, err error) { | |
| - return protectionTimestamps, (*hlc.Clock)(c).Now(), nil | |
| -} | |
| diff --git a/pkg/kv/kvserver/protectedts/protectedts_test.go b/pkg/kv/kvserver/protectedts/protectedts_test.go | |
| index 8fd0ff5149..8b3b6f8752 100644 | |
| --- a/pkg/kv/kvserver/protectedts/protectedts_test.go | |
| +++ b/pkg/kv/kvserver/protectedts/protectedts_test.go | |
| @@ -17,6 +17,7 @@ func TestProtectedTimestamps(t *testing.T) { | |
| var ( | |
| _ Provider | |
| _ Cache | |
| + _ Verifier | |
| _ Storage | |
| _ = EmptyCache(nil) | |
| _ = ErrNotExists | |
| diff --git a/pkg/kv/kvserver/protectedts/ptcache/cache.go b/pkg/kv/kvserver/protectedts/ptcache/cache.go | |
| index 6d277ffb65..b379cdf498 100644 | |
| --- a/pkg/kv/kvserver/protectedts/ptcache/cache.go | |
| +++ b/pkg/kv/kvserver/protectedts/ptcache/cache.go | |
| @@ -126,21 +126,6 @@ func (c *Cache) Refresh(ctx context.Context, asOf hlc.Timestamp) error { | |
| return nil | |
| } | |
| -// GetProtectionTimestamps is part of the spanconfig.ProtectedTSReader | |
| -// interface. | |
| -func (c *Cache) GetProtectionTimestamps( | |
| - ctx context.Context, sp roachpb.Span, | |
| -) (protectionTimestamps []hlc.Timestamp, asOf hlc.Timestamp, err error) { | |
| - readAt := c.Iterate(ctx, | |
| - sp.Key, | |
| - sp.EndKey, | |
| - func(rec *ptpb.Record) (wantMore bool) { | |
| - protectionTimestamps = append(protectionTimestamps, rec.Timestamp) | |
| - return true | |
| - }) | |
| - return protectionTimestamps, readAt, nil | |
| -} | |
| - | |
| // Start starts the periodic fetching of the Cache. A Cache must not be used | |
| // until after it has been started. An error will be returned if it has | |
| // already been started. | |
| diff --git a/pkg/kv/kvserver/protectedts/ptcache/cache_test.go b/pkg/kv/kvserver/protectedts/ptcache/cache_test.go | |
| index 37a5dd43ad..cd251dd372 100644 | |
| --- a/pkg/kv/kvserver/protectedts/ptcache/cache_test.go | |
| +++ b/pkg/kv/kvserver/protectedts/ptcache/cache_test.go | |
| @@ -68,7 +68,7 @@ func TestCacheBasic(t *testing.T) { | |
| // Then we'll add a record and make sure it gets seen. | |
| sp := tableSpan(42) | |
| - r, createdAt := protect(t, tc.Server(0), p, s.Clock().Now(), sp) | |
| + r, createdAt := protect(t, tc.Server(0), p, sp) | |
| testutils.SucceedsSoon(t, func() error { | |
| var coveredBy []*ptpb.Record | |
| seenTS := c.Iterate(ctx, sp.Key, sp.EndKey, | |
| @@ -140,7 +140,7 @@ func TestRefresh(t *testing.T) { | |
| st.verifyCounters(t, 1, 0) // just need to scan meta | |
| }) | |
| t.Run("needs refresh, with change", func(t *testing.T) { | |
| - _, createdAt := protect(t, s, p, s.Clock().Now(), metaTableSpan) | |
| + _, createdAt := protect(t, s, p, metaTableSpan) | |
| st.resetCounters() | |
| require.NoError(t, c.Refresh(ctx, createdAt)) | |
| st.verifyCounters(t, 2, 1) // need to scan meta and then scan everything | |
| @@ -177,7 +177,7 @@ func TestRefresh(t *testing.T) { | |
| require.Regexp(t, "boom", c.Refresh(ctx, s.Clock().Now()).Error()) | |
| }) | |
| t.Run("error propagates while fetching records", func(t *testing.T) { | |
| - protect(t, s, p, s.Clock().Now(), metaTableSpan) | |
| + protect(t, s, p, metaTableSpan) | |
| st.setFilter(func(ba roachpb.BatchRequest) *roachpb.Error { | |
| if scanReq, ok := ba.GetArg(roachpb.Scan); ok { | |
| scan := scanReq.(*roachpb.ScanRequest) | |
| @@ -192,7 +192,7 @@ func TestRefresh(t *testing.T) { | |
| }) | |
| t.Run("Iterate does not hold mutex", func(t *testing.T) { | |
| inIterate := make(chan chan struct{}) | |
| - rec, createdAt := protect(t, s, p, s.Clock().Now(), metaTableSpan) | |
| + rec, createdAt := protect(t, s, p, metaTableSpan) | |
| require.NoError(t, c.Refresh(ctx, createdAt)) | |
| go c.Iterate(ctx, keys.MinKey, keys.MaxKey, func(r *ptpb.Record) (wantMore bool) { | |
| if r.ID.GetUUID() != rec.ID.GetUUID() { | |
| @@ -271,8 +271,8 @@ func TestQueryRecord(t *testing.T) { | |
| waitForAsOfAfter(t, c, hlc.Timestamp{}) | |
| // Create two records. | |
| sp42 := tableSpan(42) | |
| - r1, createdAt1 := protect(t, s, p, s.Clock().Now(), sp42) | |
| - r2, createdAt2 := protect(t, s, p, s.Clock().Now(), sp42) | |
| + r1, createdAt1 := protect(t, s, p, sp42) | |
| + r2, createdAt2 := protect(t, s, p, sp42) | |
| // Ensure they both don't exist and that the read timestamps precede the | |
| // create timestamps. | |
| exists1, asOf := c.QueryRecord(ctx, r1.ID.GetUUID()) | |
| @@ -291,7 +291,7 @@ func TestQueryRecord(t *testing.T) { | |
| require.True(t, !asOf.Less(createdAt2)) | |
| // Release 2 and then create 3. | |
| require.NoError(t, p.Release(ctx, nil /* txn */, r2.ID.GetUUID())) | |
| - r3, createdAt3 := protect(t, s, p, s.Clock().Now(), sp42) | |
| + r3, createdAt3 := protect(t, s, p, sp42) | |
| exists2, asOf = c.QueryRecord(ctx, r2.ID.GetUUID()) | |
| require.True(t, exists2) | |
| require.True(t, asOf.Less(createdAt3)) | |
| @@ -329,10 +329,10 @@ func TestIterate(t *testing.T) { | |
| sp42 := tableSpan(42) | |
| sp43 := tableSpan(43) | |
| sp44 := tableSpan(44) | |
| - r1, _ := protect(t, s, p, s.Clock().Now(), sp42) | |
| - r2, _ := protect(t, s, p, s.Clock().Now(), sp43) | |
| - r3, _ := protect(t, s, p, s.Clock().Now(), sp44) | |
| - r4, _ := protect(t, s, p, s.Clock().Now(), sp42, sp43) | |
| + r1, _ := protect(t, s, p, sp42) | |
| + r2, _ := protect(t, s, p, sp43) | |
| + r3, _ := protect(t, s, p, sp44) | |
| + r4, _ := protect(t, s, p, sp42, sp43) | |
| require.NoError(t, c.Refresh(ctx, s.Clock().Now())) | |
| t.Run("all", func(t *testing.T) { | |
| var recs records | |
| @@ -373,102 +373,6 @@ func (recs *records) sorted() []*ptpb.Record { | |
| return *recs | |
| } | |
| -func TestGetProtectionTimestamps(t *testing.T) { | |
| - ctx := context.Background() | |
| - tc := testcluster.StartTestCluster(t, 1, base.TestClusterArgs{}) | |
| - defer tc.Stopper().Stop(ctx) | |
| - // Set the poll interval to be very long. | |
| - s := tc.Server(0) | |
| - protectedts.PollInterval.Override(ctx, &s.ClusterSettings().SV, 500*time.Hour) | |
| - | |
| - ts := func(nanos int) hlc.Timestamp { | |
| - return hlc.Timestamp{ | |
| - WallTime: int64(nanos), | |
| - } | |
| - } | |
| - sp42 := tableSpan(42) | |
| - sp43 := tableSpan(43) | |
| - sp44 := tableSpan(44) | |
| - sp4243 := roachpb.Span{Key: sp42.Key, EndKey: sp43.EndKey} | |
| - | |
| - for _, testCase := range []struct { | |
| - name string | |
| - test func(t *testing.T, p protectedts.Storage, c *ptcache.Cache, cleanup func(...*ptpb.Record)) | |
| - }{ | |
| - { | |
| - name: "multiple records apply to a single span", | |
| - test: func(t *testing.T, p protectedts.Storage, c *ptcache.Cache, cleanup func(...*ptpb.Record)) { | |
| - r1, _ := protect(t, s, p, ts(10), sp42) | |
| - r2, _ := protect(t, s, p, ts(11), sp42) | |
| - r3, _ := protect(t, s, p, ts(6), sp42) | |
| - require.NoError(t, c.Refresh(ctx, s.Clock().Now())) | |
| - | |
| - protectionTimestamps, _, err := c.GetProtectionTimestamps(ctx, sp42) | |
| - require.NoError(t, err) | |
| - sort.Slice(protectionTimestamps, func(i, j int) bool { | |
| - return protectionTimestamps[i].Less(protectionTimestamps[j]) | |
| - }) | |
| - require.Equal(t, []hlc.Timestamp{ts(6), ts(10), ts(11)}, protectionTimestamps) | |
| - cleanup(r1, r2, r3) | |
| - }, | |
| - }, | |
| - { | |
| - name: "no records apply", | |
| - test: func(t *testing.T, p protectedts.Storage, c *ptcache.Cache, cleanup func(...*ptpb.Record)) { | |
| - r1, _ := protect(t, s, p, ts(5), sp43) | |
| - r2, _ := protect(t, s, p, ts(10), sp44) | |
| - require.NoError(t, c.Refresh(ctx, s.Clock().Now())) | |
| - protectionTimestamps, _, err := c.GetProtectionTimestamps(ctx, sp42) | |
| - require.NoError(t, err) | |
| - require.Equal(t, []hlc.Timestamp(nil), protectionTimestamps) | |
| - cleanup(r1, r2) | |
| - }, | |
| - }, | |
| - { | |
| - name: "multiple overlapping spans multiple records", | |
| - test: func(t *testing.T, p protectedts.Storage, c *ptcache.Cache, cleanup func(...*ptpb.Record)) { | |
| - r1, _ := protect(t, s, p, ts(10), sp42) | |
| - r2, _ := protect(t, s, p, ts(15), sp42) | |
| - r3, _ := protect(t, s, p, ts(5), sp43) | |
| - r4, _ := protect(t, s, p, ts(6), sp43) | |
| - r5, _ := protect(t, s, p, ts(25), keys.EverythingSpan) | |
| - // Also add a record that doesn't overlap with the requested span and | |
| - // ensure it isn't retrieved below. | |
| - r6, _ := protect(t, s, p, ts(20), sp44) | |
| - require.NoError(t, c.Refresh(ctx, s.Clock().Now())) | |
| - | |
| - protectionTimestamps, _, err := c.GetProtectionTimestamps(ctx, sp4243) | |
| - require.NoError(t, err) | |
| - sort.Slice(protectionTimestamps, func(i, j int) bool { | |
| - return protectionTimestamps[i].Less(protectionTimestamps[j]) | |
| - }) | |
| - require.Equal( | |
| - t, []hlc.Timestamp{ts(5), ts(6), ts(10), ts(15), ts(25)}, protectionTimestamps, | |
| - ) | |
| - cleanup(r1, r2, r3, r4, r5, r6) | |
| - }, | |
| - }, | |
| - } { | |
| - t.Run(testCase.name, func(t *testing.T) { | |
| - p := ptstorage.WithDatabase(ptstorage.New(s.ClusterSettings(), | |
| - s.InternalExecutor().(sqlutil.InternalExecutor), nil /* knobs */), s.DB()) | |
| - | |
| - c := ptcache.New(ptcache.Config{ | |
| - Settings: s.ClusterSettings(), | |
| - DB: s.DB(), | |
| - Storage: p, | |
| - }) | |
| - require.NoError(t, c.Start(ctx, tc.Stopper())) | |
| - | |
| - testCase.test(t, p, c, func(records ...*ptpb.Record) { | |
| - for _, r := range records { | |
| - require.NoError(t, p.Release(ctx, nil, r.ID.GetUUID())) | |
| - } | |
| - }) | |
| - }) | |
| - } | |
| -} | |
| - | |
| func TestSettingChangedLeadsToFetch(t *testing.T) { | |
| ctx := context.Background() | |
| tc := testcluster.StartTestCluster(t, 1, base.TestClusterArgs{}) | |
| @@ -521,12 +425,9 @@ func tableSpan(tableID uint32) roachpb.Span { | |
| } | |
| func protect( | |
| - t *testing.T, | |
| - s serverutils.TestServerInterface, | |
| - p protectedts.Storage, | |
| - protectTS hlc.Timestamp, | |
| - spans ...roachpb.Span, | |
| + t *testing.T, s serverutils.TestServerInterface, p protectedts.Storage, spans ...roachpb.Span, | |
| ) (r *ptpb.Record, createdAt hlc.Timestamp) { | |
| + protectTS := s.Clock().Now() | |
| r = &ptpb.Record{ | |
| ID: uuid.MakeV4().GetBytes(), | |
| Timestamp: protectTS, | |
| diff --git a/pkg/kv/kvserver/protectedts/ptprovider/BUILD.bazel b/pkg/kv/kvserver/protectedts/ptprovider/BUILD.bazel | |
| index cd021496f2..e3620f19cb 100644 | |
| --- a/pkg/kv/kvserver/protectedts/ptprovider/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/protectedts/ptprovider/BUILD.bazel | |
| @@ -12,6 +12,7 @@ go_library( | |
| "//pkg/kv/kvserver/protectedts/ptcache", | |
| "//pkg/kv/kvserver/protectedts/ptreconcile", | |
| "//pkg/kv/kvserver/protectedts/ptstorage", | |
| + "//pkg/kv/kvserver/protectedts/ptverifier", | |
| "//pkg/settings/cluster", | |
| "//pkg/sql/sqlutil", | |
| "//pkg/util/metric", | |
| diff --git a/pkg/kv/kvserver/protectedts/ptprovider/provider.go b/pkg/kv/kvserver/protectedts/ptprovider/provider.go | |
| index 377039e80c..c537151bf2 100644 | |
| --- a/pkg/kv/kvserver/protectedts/ptprovider/provider.go | |
| +++ b/pkg/kv/kvserver/protectedts/ptprovider/provider.go | |
| @@ -21,6 +21,7 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptcache" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptreconcile" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptstorage" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptverifier" | |
| "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| "github.com/cockroachdb/cockroach/pkg/sql/sqlutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/metric" | |
| @@ -38,9 +39,9 @@ type Config struct { | |
| Knobs *protectedts.TestingKnobs | |
| } | |
| -// Provider is the concrete implementation of protectedts.Provider interface. | |
| -type Provider struct { | |
| +type provider struct { | |
| protectedts.Storage | |
| + protectedts.Verifier | |
| protectedts.Cache | |
| protectedts.Reconciler | |
| metric.Struct | |
| @@ -52,16 +53,16 @@ func New(cfg Config) (protectedts.Provider, error) { | |
| return nil, err | |
| } | |
| storage := ptstorage.New(cfg.Settings, cfg.InternalExecutor, cfg.Knobs) | |
| + verifier := ptverifier.New(cfg.DB, storage) | |
| reconciler := ptreconcile.New(cfg.Settings, cfg.DB, storage, cfg.ReconcileStatusFuncs) | |
| - cache := ptcache.New(ptcache.Config{ | |
| - DB: cfg.DB, | |
| - Storage: storage, | |
| - Settings: cfg.Settings, | |
| - }) | |
| - | |
| - return &Provider{ | |
| - Storage: storage, | |
| - Cache: cache, | |
| + return &provider{ | |
| + Storage: storage, | |
| + Cache: ptcache.New(ptcache.Config{ | |
| + DB: cfg.DB, | |
| + Storage: storage, | |
| + Settings: cfg.Settings, | |
| + }), | |
| + Verifier: verifier, | |
| Reconciler: reconciler, | |
| Struct: reconciler.Metrics(), | |
| }, nil | |
| @@ -80,15 +81,13 @@ func validateConfig(cfg Config) error { | |
| } | |
| } | |
| -// Start implements the protectedts.Provider interface. | |
| -func (p *Provider) Start(ctx context.Context, stopper *stop.Stopper) error { | |
| +func (p *provider) Start(ctx context.Context, stopper *stop.Stopper) error { | |
| if cache, ok := p.Cache.(*ptcache.Cache); ok { | |
| return cache.Start(ctx, stopper) | |
| } | |
| return nil | |
| } | |
| -// Metrics implements the protectedts.Provider interface. | |
| -func (p *Provider) Metrics() metric.Struct { | |
| +func (p *provider) Metrics() metric.Struct { | |
| return p.Struct | |
| } | |
| diff --git a/pkg/kv/kvserver/protectedts/ptstorage/BUILD.bazel b/pkg/kv/kvserver/protectedts/ptstorage/BUILD.bazel | |
| index 1879f9fa65..f94955048a 100644 | |
| --- a/pkg/kv/kvserver/protectedts/ptstorage/BUILD.bazel | |
| +++ b/pkg/kv/kvserver/protectedts/ptstorage/BUILD.bazel | |
| @@ -39,7 +39,6 @@ go_test( | |
| "validate_test.go", | |
| ], | |
| embed = [":ptstorage"], | |
| - shard_count = 16, | |
| deps = [ | |
| "//pkg/base", | |
| "//pkg/keys", | |
| diff --git a/pkg/kv/kvserver/protectedts/ptverifier/BUILD.bazel b/pkg/kv/kvserver/protectedts/ptverifier/BUILD.bazel | |
| new file mode 100644 | |
| index 0000000000..969c0cba66 | |
| --- /dev/null | |
| +++ b/pkg/kv/kvserver/protectedts/ptverifier/BUILD.bazel | |
| @@ -0,0 +1,48 @@ | |
| +load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test") | |
| + | |
| +go_library( | |
| + name = "ptverifier", | |
| + srcs = ["verifier.go"], | |
| + importpath = "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptverifier", | |
| + visibility = ["//visibility:public"], | |
| + deps = [ | |
| + "//pkg/kv", | |
| + "//pkg/kv/kvserver/protectedts", | |
| + "//pkg/kv/kvserver/protectedts/ptpb", | |
| + "//pkg/roachpb", | |
| + "//pkg/util/hlc", | |
| + "//pkg/util/uuid", | |
| + "@com_github_cockroachdb_errors//:errors", | |
| + ], | |
| +) | |
| + | |
| +go_test( | |
| + name = "ptverifier_test", | |
| + size = "small", | |
| + srcs = [ | |
| + "main_test.go", | |
| + "verifier_test.go", | |
| + ], | |
| + deps = [ | |
| + ":ptverifier", | |
| + "//pkg/base", | |
| + "//pkg/keys", | |
| + "//pkg/kv", | |
| + "//pkg/kv/kvclient/kvcoord", | |
| + "//pkg/kv/kvserver/protectedts", | |
| + "//pkg/kv/kvserver/protectedts/ptpb", | |
| + "//pkg/kv/kvserver/protectedts/ptstorage", | |
| + "//pkg/roachpb", | |
| + "//pkg/security", | |
| + "//pkg/security/securitytest", | |
| + "//pkg/server", | |
| + "//pkg/sql/sqlutil", | |
| + "//pkg/testutils/serverutils", | |
| + "//pkg/testutils/testcluster", | |
| + "//pkg/util/leaktest", | |
| + "//pkg/util/randutil", | |
| + "//pkg/util/uuid", | |
| + "@com_github_cockroachdb_errors//:errors", | |
| + "@com_github_stretchr_testify//require", | |
| + ], | |
| +) | |
| diff --git a/pkg/sql/scheduledlogging/main_test.go b/pkg/kv/kvserver/protectedts/ptverifier/main_test.go | |
| similarity index 93% | |
| rename from pkg/sql/scheduledlogging/main_test.go | |
| rename to pkg/kv/kvserver/protectedts/ptverifier/main_test.go | |
| index 6f02005f4e..8fff74647b 100644 | |
| --- a/pkg/sql/scheduledlogging/main_test.go | |
| +++ b/pkg/kv/kvserver/protectedts/ptverifier/main_test.go | |
| @@ -1,4 +1,4 @@ | |
| -// Copyright 2022 The Cockroach Authors. | |
| +// Copyright 2019 The Cockroach Authors. | |
| // | |
| // Use of this software is governed by the Business Source License | |
| // included in the file licenses/BSL.txt. | |
| @@ -8,7 +8,7 @@ | |
| // by the Apache License, Version 2.0, included in the file | |
| // licenses/APL.txt. | |
| -package scheduledlogging_test | |
| +package ptverifier_test | |
| import ( | |
| "os" | |
| diff --git a/pkg/kv/kvserver/protectedts/ptverifier/verifier.go b/pkg/kv/kvserver/protectedts/ptverifier/verifier.go | |
| new file mode 100644 | |
| index 0000000000..c3f46349c8 | |
| --- /dev/null | |
| +++ b/pkg/kv/kvserver/protectedts/ptverifier/verifier.go | |
| @@ -0,0 +1,147 @@ | |
| +// Copyright 2019 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package ptverifier | |
| + | |
| +import ( | |
| + "bytes" | |
| + "context" | |
| + "fmt" | |
| + | |
| + "github.com/cockroachdb/cockroach/pkg/kv" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptpb" | |
| + "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| + "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| + "github.com/cockroachdb/cockroach/pkg/util/uuid" | |
| + "github.com/cockroachdb/errors" | |
| +) | |
| + | |
| +// verifier implements protectedts.Verifier. | |
| +type verifier struct { | |
| + db *kv.DB | |
| + s protectedts.Storage | |
| +} | |
| + | |
| +// New returns a new Verifier. | |
| +func New(db *kv.DB, s protectedts.Storage) protectedts.Verifier { | |
| + return &verifier{db: db, s: s} | |
| +} | |
| + | |
| +// Verify verifies that a record with the provided id is verified. | |
| +// If it is not verified this call will perform verification and mark the | |
| +// record as verified. | |
| +func (v *verifier) Verify(ctx context.Context, id uuid.UUID) error { | |
| + // First we go read the record and note the timestamp at which we read it. | |
| + r, ts, err := getRecordWithTimestamp(ctx, v.s, v.db, id) | |
| + if err != nil { | |
| + return errors.Wrapf(err, "failed to fetch record %s", id) | |
| + } | |
| + | |
| + // TODO(adityamaru): Remove this once we delete all `Verify` calls. The new | |
| + // subsystem is not going to provide verification semantics. Until then mark | |
| + // the record as verified so it is a noop. | |
| + if r.DeprecatedSpans == nil { | |
| + return nil | |
| + } | |
| + | |
| + if r.Verified { // already verified | |
| + return nil | |
| + } | |
| + | |
| + b := makeVerificationBatch(r, ts) | |
| + if err := v.db.Run(ctx, &b); err != nil { | |
| + return err | |
| + } | |
| + | |
| + // Check the responses and synthesize an error if one occurred. | |
| + if err := parseResponse(&b, r); err != nil { | |
| + return err | |
| + } | |
| + // Mark the record as verified. | |
| + return errors.Wrapf(v.db.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| + return v.s.MarkVerified(ctx, txn, id) | |
| + }), "failed to mark %v as verified", id) | |
| +} | |
| + | |
| +// getRecordWithTimestamp fetches the record with the provided id and returns | |
| +// the hlc timestamp at which that read occurred. | |
| +func getRecordWithTimestamp( | |
| + ctx context.Context, s protectedts.Storage, db *kv.DB, id uuid.UUID, | |
| +) (r *ptpb.Record, readAt hlc.Timestamp, err error) { | |
| + if err = db.Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| + r, err = s.GetRecord(ctx, txn, id) | |
| + readAt = txn.ReadTimestamp() | |
| + return err | |
| + }); err != nil { | |
| + return nil, hlc.Timestamp{}, err | |
| + } | |
| + return r, readAt, nil | |
| +} | |
| + | |
| +func makeVerificationBatch(r *ptpb.Record, aliveAt hlc.Timestamp) kv.Batch { | |
| + // Need to perform validation, build a batch and run it. | |
| + mergedSpans, _ := roachpb.MergeSpans(&r.DeprecatedSpans) | |
| + var b kv.Batch | |
| + for _, s := range mergedSpans { | |
| + var req roachpb.AdminVerifyProtectedTimestampRequest | |
| + req.RecordAliveAt = aliveAt | |
| + req.Protected = r.Timestamp | |
| + req.RecordID = r.ID.GetUUID() | |
| + req.Key = s.Key | |
| + req.EndKey = s.EndKey | |
| + b.AddRawRequest(&req) | |
| + } | |
| + return b | |
| +} | |
| + | |
| +func parseResponse(b *kv.Batch, r *ptpb.Record) error { | |
| + rawResponse := b.RawResponse() | |
| + var errBuilder bytes.Buffer | |
| + for _, resp := range rawResponse.Responses { | |
| + resp := resp.GetInner().(*roachpb.AdminVerifyProtectedTimestampResponse) | |
| + if len(resp.DeprecatedFailedRanges) == 0 && len(resp.VerificationFailedRanges) == 0 { | |
| + continue | |
| + } | |
| + | |
| + // Write the error header the first time we encounter failed ranges. | |
| + if errBuilder.Len() == 0 { | |
| + _, _ = errBuilder.WriteString(fmt.Sprintf("failed to verify protection record %s with ts: %s:\n", | |
| + r.ID.String(), r.Timestamp.String())) | |
| + } | |
| + | |
| + useDeprecated := len(resp.VerificationFailedRanges) == 0 | |
| + for _, failedRange := range resp.VerificationFailedRanges { | |
| + if failedRange.Reason != "" { | |
| + // Write the per range reason for failure. | |
| + _, _ = errBuilder.WriteString(fmt.Sprintf("range ID: %d, range span: %s - %s: %s\n", | |
| + failedRange.RangeID, failedRange.StartKey.String(), failedRange.EndKey.String(), | |
| + failedRange.Reason)) | |
| + } else { | |
| + // If no reason was saved, dump relevant information. | |
| + _, _ = errBuilder.WriteString(fmt.Sprintf("range ID: %d, range span: %s - %s\n", | |
| + failedRange.RangeID, failedRange.StartKey.String(), failedRange.EndKey.String())) | |
| + } | |
| + } | |
| + | |
| + if !useDeprecated { | |
| + continue | |
| + } | |
| + | |
| + for _, rangeDesc := range resp.DeprecatedFailedRanges { | |
| + _, _ = errBuilder.WriteString(fmt.Sprintf("range ID: %d, range span: %s - %s\n", | |
| + rangeDesc.RangeID, rangeDesc.StartKey.String(), rangeDesc.EndKey.String())) | |
| + } | |
| + } | |
| + if errBuilder.Len() > 0 { | |
| + return errors.Newf("protected ts verification error: %s", errBuilder.String()) | |
| + } | |
| + return nil | |
| +} | |
| diff --git a/pkg/kv/kvserver/protectedts/ptverifier/verifier_test.go b/pkg/kv/kvserver/protectedts/ptverifier/verifier_test.go | |
| new file mode 100644 | |
| index 0000000000..de87dfbd53 | |
| --- /dev/null | |
| +++ b/pkg/kv/kvserver/protectedts/ptverifier/verifier_test.go | |
| @@ -0,0 +1,259 @@ | |
| +// Copyright 2019 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package ptverifier_test | |
| + | |
| +import ( | |
| + "context" | |
| + "sync/atomic" | |
| + "testing" | |
| + "time" | |
| + | |
| + "github.com/cockroachdb/cockroach/pkg/base" | |
| + "github.com/cockroachdb/cockroach/pkg/keys" | |
| + "github.com/cockroachdb/cockroach/pkg/kv" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvclient/kvcoord" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptpb" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptstorage" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptverifier" | |
| + "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| + "github.com/cockroachdb/cockroach/pkg/sql/sqlutil" | |
| + "github.com/cockroachdb/cockroach/pkg/testutils/testcluster" | |
| + "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| + "github.com/cockroachdb/cockroach/pkg/util/uuid" | |
| + "github.com/cockroachdb/errors" | |
| + "github.com/stretchr/testify/require" | |
| +) | |
| + | |
| +// TestVerifier tests the business logic of verification by mocking out the | |
| +// actual verification requests but using a real implementation of | |
| +// protectedts.Storage. | |
| +func TestVerifier(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + | |
| + ctx := context.Background() | |
| + tc := testcluster.StartTestCluster(t, 1, base.TestClusterArgs{}) | |
| + defer tc.Stopper().Stop(ctx) | |
| + | |
| + s := tc.Server(0) | |
| + var senderFunc atomic.Value | |
| + senderFunc.Store(kv.SenderFunc(nil)) | |
| + ds := s.DistSenderI().(*kvcoord.DistSender) | |
| + tsf := kvcoord.NewTxnCoordSenderFactory( | |
| + kvcoord.TxnCoordSenderFactoryConfig{ | |
| + AmbientCtx: s.DB().AmbientContext, | |
| + HeartbeatInterval: time.Second, | |
| + Settings: s.ClusterSettings(), | |
| + Clock: s.Clock(), | |
| + Stopper: s.Stopper(), | |
| + }, | |
| + kv.SenderFunc(func(ctx context.Context, ba roachpb.BatchRequest) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if f := senderFunc.Load().(kv.SenderFunc); f != nil { | |
| + return f(ctx, ba) | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + }), | |
| + ) | |
| + | |
| + pts := ptstorage.New(s.ClusterSettings(), s.InternalExecutor().(sqlutil.InternalExecutor), nil /* knobs */) | |
| + withDB := ptstorage.WithDatabase(pts, s.DB()) | |
| + db := kv.NewDB(s.DB().AmbientContext, tsf, s.Clock(), s.Stopper()) | |
| + ptv := ptverifier.New(db, pts) | |
| + makeTableSpan := func(tableID uint32) roachpb.Span { | |
| + k := keys.SystemSQLCodec.TablePrefix(tableID) | |
| + return roachpb.Span{Key: k, EndKey: k.PrefixEnd()} | |
| + } | |
| + | |
| + createRecord := func(t *testing.T, tables ...uint32) *ptpb.Record { | |
| + spans := make([]roachpb.Span, len(tables)) | |
| + for i, tid := range tables { | |
| + spans[i] = makeTableSpan(tid) | |
| + } | |
| + r := ptpb.Record{ | |
| + ID: uuid.MakeV4().GetBytes(), | |
| + Timestamp: s.Clock().Now(), | |
| + Mode: ptpb.PROTECT_AFTER, | |
| + DeprecatedSpans: spans, | |
| + } | |
| + require.Nil(t, s.DB().Txn(ctx, func(ctx context.Context, txn *kv.Txn) error { | |
| + return pts.Protect(ctx, txn, &r) | |
| + })) | |
| + return &r | |
| + } | |
| + ensureVerified := func(t *testing.T, id uuid.UUID, verified bool) { | |
| + got, err := withDB.GetRecord(ctx, nil, id) | |
| + require.NoError(t, err) | |
| + require.Equal(t, verified, got.Verified) | |
| + } | |
| + release := func(t *testing.T, id uuid.UUID) { | |
| + require.NoError(t, withDB.Release(ctx, nil, id)) | |
| + } | |
| + for _, c := range []struct { | |
| + name string | |
| + test func(t *testing.T) | |
| + }{ | |
| + { | |
| + name: "record doesn't exist", | |
| + test: func(t *testing.T) { | |
| + require.Regexp(t, protectedts.ErrNotExists.Error(), | |
| + ptv.Verify(ctx, uuid.MakeV4()).Error()) | |
| + }, | |
| + }, | |
| + { | |
| + name: "verification failed with injected error", | |
| + test: func(t *testing.T) { | |
| + defer senderFunc.Store(senderFunc.Load()) | |
| + r := createRecord(t, 42) | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + return nil, roachpb.NewError(errors.New("boom")) | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.Regexp(t, "boom", ptv.Verify(ctx, r.ID.GetUUID()).Error()) | |
| + ensureVerified(t, r.ID.GetUUID(), false) | |
| + release(t, r.ID.GetUUID()) | |
| + }, | |
| + }, | |
| + { | |
| + name: "verification failed with injected response", | |
| + test: func(t *testing.T) { | |
| + defer senderFunc.Store(senderFunc.Load()) | |
| + r := createRecord(t, 42) | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + var resp roachpb.BatchResponse | |
| + resp.Add(&roachpb.AdminVerifyProtectedTimestampResponse{ | |
| + VerificationFailedRanges: []roachpb.AdminVerifyProtectedTimestampResponse_FailedRange{{ | |
| + RangeID: 42, | |
| + StartKey: roachpb.RKey(r.DeprecatedSpans[0].Key), | |
| + EndKey: roachpb.RKey(r.DeprecatedSpans[0].EndKey), | |
| + }}, | |
| + }) | |
| + return &resp, nil | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.Regexp(t, "protected ts verification error: failed to verify protection.*\n"+ | |
| + "range ID: 42, range span: /Table/42 - /Table/43", | |
| + ptv.Verify(ctx, r.ID.GetUUID()).Error()) | |
| + ensureVerified(t, r.ID.GetUUID(), false) | |
| + release(t, r.ID.GetUUID()) | |
| + }, | |
| + }, | |
| + { | |
| + name: "verification failed with injected response over two spans", | |
| + test: func(t *testing.T) { | |
| + defer senderFunc.Store(senderFunc.Load()) | |
| + r := createRecord(t, 42, 12) | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + var resp roachpb.BatchResponse | |
| + resp.Add(&roachpb.AdminVerifyProtectedTimestampResponse{ | |
| + VerificationFailedRanges: []roachpb.AdminVerifyProtectedTimestampResponse_FailedRange{{ | |
| + RangeID: 42, | |
| + StartKey: roachpb.RKey(r.DeprecatedSpans[0].Key), | |
| + EndKey: roachpb.RKey(r.DeprecatedSpans[0].EndKey), | |
| + Reason: "foo", | |
| + }}, | |
| + }) | |
| + resp.Add(&roachpb.AdminVerifyProtectedTimestampResponse{ | |
| + VerificationFailedRanges: []roachpb.AdminVerifyProtectedTimestampResponse_FailedRange{{ | |
| + RangeID: 12, | |
| + StartKey: roachpb.RKey(r.DeprecatedSpans[1].Key), | |
| + EndKey: roachpb.RKey(r.DeprecatedSpans[1].EndKey), | |
| + Reason: "bar", | |
| + }}, | |
| + }) | |
| + return &resp, nil | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.Regexp(t, "protected ts verification error: failed to verify protection.*\n"+ | |
| + "range ID: 42, "+ | |
| + "range span: /Table/42 - /Table/43: foo\nrange ID: 12, "+ | |
| + "range span: /Table/12 - /Table/13: bar", | |
| + ptv.Verify(ctx, r.ID.GetUUID()).Error()) | |
| + ensureVerified(t, r.ID.GetUUID(), false) | |
| + release(t, r.ID.GetUUID()) | |
| + }, | |
| + }, | |
| + { | |
| + // TODO(adityamaru): Remove in 21.2. | |
| + name: "verification failed with deprecated failed ranges response", | |
| + test: func(t *testing.T) { | |
| + defer senderFunc.Store(senderFunc.Load()) | |
| + r := createRecord(t, 42) | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + var resp roachpb.BatchResponse | |
| + resp.Add(&roachpb.AdminVerifyProtectedTimestampResponse{ | |
| + DeprecatedFailedRanges: []roachpb.RangeDescriptor{{ | |
| + RangeID: 42, | |
| + StartKey: roachpb.RKey(r.DeprecatedSpans[0].Key), | |
| + EndKey: roachpb.RKey(r.DeprecatedSpans[0].EndKey), | |
| + }}, | |
| + }) | |
| + return &resp, nil | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.Regexp(t, "protected ts verification error: failed to verify protection."+ | |
| + "*\nrange ID: 42, range span: /Table/42 - /Table/43", | |
| + ptv.Verify(ctx, r.ID.GetUUID()).Error()) | |
| + ensureVerified(t, r.ID.GetUUID(), false) | |
| + release(t, r.ID.GetUUID()) | |
| + }, | |
| + }, | |
| + { | |
| + name: "verification succeeded", | |
| + test: func(t *testing.T) { | |
| + defer senderFunc.Store(senderFunc.Load()) | |
| + r := createRecord(t, 42) | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + var resp roachpb.BatchResponse | |
| + resp.Add(&roachpb.AdminVerifyProtectedTimestampResponse{}) | |
| + return &resp, nil | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.NoError(t, ptv.Verify(ctx, r.ID.GetUUID())) | |
| + ensureVerified(t, r.ID.GetUUID(), true) | |
| + // Show that we don't send again once we've already verified. | |
| + sawVerification := false | |
| + senderFunc.Store(kv.SenderFunc(func( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| + ) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + if _, ok := ba.GetArg(roachpb.AdminVerifyProtectedTimestamp); ok { | |
| + sawVerification = true | |
| + } | |
| + return ds.Send(ctx, ba) | |
| + })) | |
| + require.NoError(t, ptv.Verify(ctx, r.ID.GetUUID())) | |
| + require.False(t, sawVerification) | |
| + release(t, r.ID.GetUUID()) | |
| + }, | |
| + }, | |
| + } { | |
| + t.Run(c.name, c.test) | |
| + } | |
| +} | |
| diff --git a/pkg/kv/kvserver/queue.go b/pkg/kv/kvserver/queue.go | |
| index e9adac48e1..40034f86a9 100644 | |
| --- a/pkg/kv/kvserver/queue.go | |
| +++ b/pkg/kv/kvserver/queue.go | |
| @@ -33,7 +33,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| ) | |
| const ( | |
| @@ -569,7 +568,7 @@ func (bq *baseQueue) Async( | |
| ctx context.Context, opName string, wait bool, fn func(ctx context.Context, h queueHelper), | |
| ) { | |
| if log.V(3) { | |
| - log.InfofDepth(ctx, 2, "%s", redact.Safe(opName)) | |
| + log.InfofDepth(ctx, 2, "%s", log.Safe(opName)) | |
| } | |
| opName += " (" + bq.name + ")" | |
| bgCtx := bq.AnnotateCtx(context.Background()) | |
| @@ -582,7 +581,7 @@ func (bq *baseQueue) Async( | |
| func(ctx context.Context) { | |
| fn(ctx, baseQueueHelper{bq}) | |
| }); err != nil && bq.addLogN.ShouldLog() { | |
| - log.Infof(ctx, "rate limited in %s: %s", redact.Safe(opName), err) | |
| + log.Infof(ctx, "rate limited in %s: %s", log.Safe(opName), err) | |
| } | |
| } | |
| diff --git a/pkg/kv/kvserver/raft.go b/pkg/kv/kvserver/raft.go | |
| index a60f73ddfb..6d71515ab9 100644 | |
| --- a/pkg/kv/kvserver/raft.go | |
| +++ b/pkg/kv/kvserver/raft.go | |
| @@ -19,7 +19,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverbase" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverpb" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| - "github.com/cockroachdb/redact" | |
| "go.etcd.io/etcd/raft/v3" | |
| "go.etcd.io/etcd/raft/v3/raftpb" | |
| ) | |
| @@ -121,29 +120,29 @@ func wrapNumbersAsSafe(v ...interface{}) { | |
| for i := range v { | |
| switch v[i].(type) { | |
| case uint: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case uint8: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case uint16: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case uint32: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case uint64: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case int: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case int8: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case int16: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case int32: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case int64: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case float32: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| case float64: | |
| - v[i] = redact.Safe(v[i]) | |
| + v[i] = log.Safe(v[i]) | |
| default: | |
| } | |
| } | |
| diff --git a/pkg/kv/kvserver/raft_log_queue.go b/pkg/kv/kvserver/raft_log_queue.go | |
| index 408009389d..9cf07ad296 100644 | |
| --- a/pkg/kv/kvserver/raft_log_queue.go | |
| +++ b/pkg/kv/kvserver/raft_log_queue.go | |
| @@ -28,7 +28,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/errors" | |
| - "github.com/cockroachdb/redact" | |
| "go.etcd.io/etcd/raft/v3" | |
| "go.etcd.io/etcd/raft/v3/tracker" | |
| ) | |
| @@ -364,17 +363,6 @@ const ( | |
| truncatableIndexChosenViaLastIndex = "last index" | |
| ) | |
| -// No assumption should be made about the relationship between | |
| -// RaftStatus.Commit, FirstIndex, LastIndex. This is because: | |
| -// - In some cases they are not updated or read atomically. | |
| -// - FirstIndex is a potentially future first index, after the pending | |
| -// truncations have been applied. Currently, pending truncations are being | |
| -// proposed through raft, so one can be sure that these pending truncations | |
| -// do not refer to entries that are not already in the log. However, this | |
| -// situation may change in the future. In general, we should not make an | |
| -// assumption on what is in the local raft log based solely on FirstIndex, | |
| -// and should be based on whether [FirstIndex,LastIndex] is a non-empty | |
| -// interval. | |
| type truncateDecisionInput struct { | |
| RaftStatus raft.Status | |
| LogSize, MaxLogSize int64 | |
| @@ -580,24 +568,20 @@ func computeTruncateDecision(input truncateDecisionInput) truncateDecision { | |
| } | |
| // We've inherited the unfortunate semantics for {First,Last}Index from | |
| - // raft.Storage: both {First,Last}Index are inclusive. The way we've | |
| - // initialized repl.FirstIndex is to set it to the first index in the | |
| - // possibly-empty log (TruncatedState.Index + 1), and allowing LastIndex to | |
| - // fall behind it when the log is empty (TruncatedState.Index). The | |
| - // initialization is done when minting a new replica from either the | |
| - // truncated state of incoming snapshot, or using the default initial log | |
| - // index. This makes for the confusing situation where FirstIndex > | |
| - // LastIndex. We can detect this special empty log case by comparing | |
| - // checking if `FirstIndex == LastIndex + 1`. Similar to this, we can have | |
| - // the case that `FirstIndex = CommitIndex + 1` when there are no committed | |
| - // entries. Additionally, FirstIndex adjusts for the pending log | |
| - // truncations, which allows for FirstIndex to be greater than LastIndex and | |
| - // commited index by more than 1 (see the comment with | |
| - // truncateDecisionInput). So all invariant checking below is gated on first | |
| - // ensuring that the log is not empty, i.e., FirstIndex <= LastIndex. | |
| - // | |
| - // If the raft log is not empty, and there are committed entries, we can | |
| - // assert on the following invariants: | |
| + // raft.Storage. Specifically, both {First,Last}Index are inclusive, so | |
| + // there's no way to represent an empty log. The way we've initialized | |
| + // repl.FirstIndex is to set it to the first index in the possibly-empty log | |
| + // (TruncatedState.Index + 1), and allowing LastIndex to fall behind it when | |
| + // the log is empty (TruncatedState.Index). The initialization is done when | |
| + // minting a new replica from either the truncated state of incoming | |
| + // snapshot, or using the default initial log index. This makes for the | |
| + // confusing situation where FirstIndex > LastIndex. We can detect this | |
| + // special empty log case by comparing checking if | |
| + // `FirstIndex == LastIndex + 1` (`logEmpty` below). Similar to this, we can | |
| + // have the case that `FirstIndex = CommitIndex + 1` when there are no | |
| + // committed entries (which we check for in `noCommittedEntries` below). | |
| + // Having done that (i.e. if the raft log is not empty, and there are | |
| + // committed entries), we can assert on the following invariants: | |
| // | |
| // FirstIndex <= LastIndex (0) | |
| // NewFirstIndex >= FirstIndex (1) | |
| @@ -617,8 +601,8 @@ func computeTruncateDecision(input truncateDecisionInput) truncateDecision { | |
| // consider the empty case. Something like | |
| // https://github.com/nvanbenschoten/optional could help us emulate an | |
| // `option<uint64>` type if we care enough. | |
| - logEmpty := input.FirstIndex > input.LastIndex | |
| - noCommittedEntries := input.FirstIndex > input.RaftStatus.Commit | |
| + logEmpty := input.FirstIndex == input.LastIndex+1 | |
| + noCommittedEntries := input.FirstIndex == input.RaftStatus.Commit+1 | |
| logIndexValid := logEmpty || | |
| (decision.NewFirstIndex >= input.FirstIndex) && (decision.NewFirstIndex <= input.LastIndex) | |
| @@ -720,14 +704,14 @@ func (rlq *raftLogQueue) process( | |
| // Can and should the raft logs be truncated? | |
| if !decision.ShouldTruncate() { | |
| - log.VEventf(ctx, 3, "%s", redact.Safe(decision.String())) | |
| + log.VEventf(ctx, 3, "%s", log.Safe(decision.String())) | |
| return false, nil | |
| } | |
| if n := decision.NumNewRaftSnapshots(); log.V(1) || n > 0 && rlq.logSnapshots.ShouldProcess(timeutil.Now()) { | |
| - log.Infof(ctx, "%v", redact.Safe(decision.String())) | |
| + log.Infof(ctx, "%v", log.Safe(decision.String())) | |
| } else { | |
| - log.VEventf(ctx, 1, "%v", redact.Safe(decision.String())) | |
| + log.VEventf(ctx, 1, "%v", log.Safe(decision.String())) | |
| } | |
| b := &kv.Batch{} | |
| truncRequest := &roachpb.TruncateLogRequest{ | |
| diff --git a/pkg/kv/kvserver/raft_log_queue_test.go b/pkg/kv/kvserver/raft_log_queue_test.go | |
| index d2ba4982f4..20adf8fad7 100644 | |
| --- a/pkg/kv/kvserver/raft_log_queue_test.go | |
| +++ b/pkg/kv/kvserver/raft_log_queue_test.go | |
| @@ -925,19 +925,12 @@ func waitForTruncationForTesting( | |
| // Flush the engine to advance durability, which triggers truncation. | |
| require.NoError(t, r.Engine().Flush()) | |
| } | |
| - // FirstIndex should have changed. | |
| + // FirstIndex has changed. | |
| firstIndex, err := r.GetFirstIndex() | |
| require.NoError(t, err) | |
| if firstIndex != newFirstIndex { | |
| return errors.Errorf("expected firstIndex == %d, got %d", newFirstIndex, firstIndex) | |
| } | |
| - // Some low-level tests also look at the raftEntryCache or sideloaded | |
| - // storage, which are updated after, and non-atomically with the change to | |
| - // first index (latter holds Replica.mu). Since the raftLogTruncator holds Replica.raftMu | |
| - // for the duration of its work, we can, by acquiring and releasing raftMu here, ensure | |
| - // that we have waited for it to finish. | |
| - r.raftMu.Lock() | |
| - defer r.raftMu.Unlock() | |
| return nil | |
| }) | |
| } | |
| diff --git a/pkg/kv/kvserver/raft_log_truncator.go b/pkg/kv/kvserver/raft_log_truncator.go | |
| index 47c34d7e61..ad0aef0c48 100644 | |
| --- a/pkg/kv/kvserver/raft_log_truncator.go | |
| +++ b/pkg/kv/kvserver/raft_log_truncator.go | |
| @@ -380,10 +380,19 @@ func (r rangesByRangeID) Swap(i, j int) { | |
| // deadlock (see storage.Engine.RegisterFlushCompletedCallback). | |
| func (t *raftLogTruncator) durabilityAdvancedCallback() { | |
| runTruncation := false | |
| + doneRunning := func() {} | |
| t.mu.Lock() | |
| if !t.mu.runningTruncation && len(t.mu.addRanges) > 0 { | |
| runTruncation = true | |
| t.mu.runningTruncation = true | |
| + doneRunning = func() { | |
| + t.mu.Lock() | |
| + defer t.mu.Unlock() | |
| + if !t.mu.runningTruncation { | |
| + panic("expected runningTruncation") | |
| + } | |
| + t.mu.runningTruncation = false | |
| + } | |
| } | |
| if !runTruncation && len(t.mu.addRanges) > 0 { | |
| t.mu.queuedDurabilityCB = true | |
| @@ -394,31 +403,20 @@ func (t *raftLogTruncator) durabilityAdvancedCallback() { | |
| } | |
| if err := t.stopper.RunAsyncTask(t.ambientCtx, "raft-log-truncation", | |
| func(ctx context.Context) { | |
| + defer doneRunning() | |
| for { | |
| t.durabilityAdvanced(ctx) | |
| - shouldReturn := false | |
| t.mu.Lock() | |
| queued := t.mu.queuedDurabilityCB | |
| t.mu.queuedDurabilityCB = false | |
| - if !queued { | |
| - t.mu.runningTruncation = false | |
| - shouldReturn = true | |
| - } | |
| t.mu.Unlock() | |
| - if shouldReturn { | |
| + if !queued { | |
| return | |
| } | |
| } | |
| }); err != nil { | |
| // Task did not run because stopper is stopped. | |
| - func() { | |
| - t.mu.Lock() | |
| - defer t.mu.Unlock() | |
| - if !t.mu.runningTruncation { | |
| - panic("expected runningTruncation") | |
| - } | |
| - t.mu.runningTruncation = false | |
| - }() | |
| + doneRunning() | |
| } | |
| } | |
| diff --git a/pkg/kv/kvserver/replica.go b/pkg/kv/kvserver/replica.go | |
| index f1e4863662..65a593eabf 100644 | |
| --- a/pkg/kv/kvserver/replica.go | |
| +++ b/pkg/kv/kvserver/replica.go | |
| @@ -1024,8 +1024,8 @@ func (r *Replica) getImpliedGCThresholdRLocked( | |
| // If we have a protected timestamp record which precedes the implied | |
| // threshold, use the threshold it implies instead. | |
| - if !c.earliestProtectionTimestamp.IsEmpty() && c.earliestProtectionTimestamp.Less(threshold) { | |
| - return c.earliestProtectionTimestamp.Prev() | |
| + if c.earliestRecord != nil && c.earliestRecord.Timestamp.Less(threshold) { | |
| + return c.earliestRecord.Timestamp.Prev() | |
| } | |
| return threshold | |
| } | |
| @@ -1303,7 +1303,7 @@ func (r *Replica) assertStateRaftMuLockedReplicaMuRLocked( | |
| pretty.Diff(diskState, r.mu.state)) | |
| r.mu.state.Desc, diskState.Desc = nil, nil | |
| log.Fatalf(ctx, "on-disk and in-memory state diverged: %s", | |
| - redact.Safe(pretty.Diff(diskState, r.mu.state))) | |
| + log.Safe(pretty.Diff(diskState, r.mu.state))) | |
| } | |
| if r.isInitializedRLocked() { | |
| if !r.startKey.Equal(r.mu.state.Desc.StartKey) { | |
| diff --git a/pkg/kv/kvserver/replica_application_cmd.go b/pkg/kv/kvserver/replica_application_cmd.go | |
| index 8d738a77ba..07c056babe 100644 | |
| --- a/pkg/kv/kvserver/replica_application_cmd.go | |
| +++ b/pkg/kv/kvserver/replica_application_cmd.go | |
| @@ -122,7 +122,9 @@ func (c *replicatedCmd) Ctx() context.Context { | |
| // AckErrAndFinish implements the apply.Command interface. | |
| func (c *replicatedCmd) AckErrAndFinish(ctx context.Context, err error) error { | |
| if c.IsLocal() { | |
| - c.response.Err = roachpb.NewError(roachpb.NewAmbiguousResultError(err)) | |
| + c.response.Err = roachpb.NewError( | |
| + roachpb.NewAmbiguousResultError( | |
| + err.Error())) | |
| } | |
| return c.AckOutcomeAndFinish(ctx) | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_backpressure.go b/pkg/kv/kvserver/replica_backpressure.go | |
| index 8ac0388794..3ae46e823c 100644 | |
| --- a/pkg/kv/kvserver/replica_backpressure.go | |
| +++ b/pkg/kv/kvserver/replica_backpressure.go | |
| @@ -108,19 +108,17 @@ func canBackpressureBatch(ba *roachpb.BatchRequest) bool { | |
| return false | |
| } | |
| -// signallerForBatch returns the signaller to use for this batch. This is the | |
| -// Replica's breaker's signaller except if any request in the batch uses | |
| -// poison.Policy_Wait, in which case it's a neverTripSignaller. In particular, | |
| -// `(signaller).C() == nil` signals that the request bypasses the circuit | |
| -// breakers. | |
| -func (r *Replica) signallerForBatch(ba *roachpb.BatchRequest) signaller { | |
| +// bypassReplicaCircuitBreakerForBatch returns whether the provided | |
| +// BatchRequest bypasses the per-Replica circuit breaker. This is the | |
| +// case if any request in the batch is requesting to do so. | |
| +func bypassReplicaCircuitBreakerForBatch(ba *roachpb.BatchRequest) bool { | |
| for _, ru := range ba.Requests { | |
| req := ru.GetInner() | |
| if roachpb.BypassesReplicaCircuitBreaker(req) { | |
| - return neverTripSignaller{} | |
| + return true | |
| } | |
| } | |
| - return r.breaker.Signal() | |
| + return false | |
| } | |
| // shouldBackpressureWrites returns whether writes to the range should be | |
| diff --git a/pkg/kv/kvserver/replica_circuit_breaker.go b/pkg/kv/kvserver/replica_circuit_breaker.go | |
| index 300f1db2fd..e64ddb75f3 100644 | |
| --- a/pkg/kv/kvserver/replica_circuit_breaker.go | |
| +++ b/pkg/kv/kvserver/replica_circuit_breaker.go | |
| @@ -37,8 +37,7 @@ type replicaInCircuitBreaker interface { | |
| Desc() *roachpb.RangeDescriptor | |
| Send(context.Context, roachpb.BatchRequest) (*roachpb.BatchResponse, *roachpb.Error) | |
| slowReplicationThreshold(ba *roachpb.BatchRequest) (time.Duration, bool) | |
| - replicaUnavailableError(err error) error | |
| - poisonInflightLatches(err error) | |
| + replicaUnavailableError() error | |
| } | |
| var defaultReplicaCircuitBreakerSlowReplicationThreshold = envutil.EnvOrDefaultDuration( | |
| @@ -75,15 +74,127 @@ type replicaCircuitBreaker struct { | |
| stopper *stop.Stopper | |
| r replicaInCircuitBreaker | |
| st *cluster.Settings | |
| + cancels CancelStorage | |
| wrapped *circuit.Breaker | |
| versionIsActive int32 // atomic | |
| } | |
| +// Register takes a cancelable context and its cancel function (which the caller | |
| +// must cancel when the request has finished), and registers them with the | |
| +// circuit breaker. If the breaker is already tripped, its error is returned | |
| +// immediately and the caller should not continue processing the request. | |
| +// Otherwise, the cancel function is invoked if the breaker trips. The caller is | |
| +// provided with a token and signaller for use in a call to | |
| +// UnregisterAndAdjustError upon request completion. That method also takes the | |
| +// error (if any) resulting from the request to ensure that in the case of a | |
| +// tripped breaker, the error reflects this fact. | |
| +func (br *replicaCircuitBreaker) Register( | |
| + ctx context.Context, cancel func(), | |
| +) (_token interface{}, _ signaller, _ error) { | |
| + brSig := br.Signal() | |
| + | |
| + // TODO(tbg): we may want to exclude more requests from this check, or allow | |
| + // requests to exclude themselves from the check (via their header). This | |
| + // latter mechanism could also replace hasBypassCircuitBreakerMarker. | |
| + if hasBypassCircuitBreakerMarker(ctx) { | |
| + // NB: brSig.C() == nil. | |
| + brSig = neverTripSignaller{} | |
| + } | |
| + | |
| + if brSig.C() == nil { | |
| + // Circuit breakers are disabled and/or this is a probe request, so don't do | |
| + // any work registering the context. UnregisterAndAdjustError will know that we didn't | |
| + // since it checks the same brSig for a nil C(). | |
| + return ctx, brSig, nil | |
| + } | |
| + | |
| + // NB: it might be tempting to check the breaker error first to avoid the call | |
| + // to Set below if the breaker is tripped at this point. However, the ordering | |
| + // here, subtly, is required to avoid situations in which the cancel is still | |
| + // in the map despite the probe having shut down (in which case cancel will | |
| + // not be invoked until the probe is next triggered, which maybe "never"). | |
| + // | |
| + // To see this, consider the case in which the breaker is initially not | |
| + // tripped when we check, but then trips immediately and has the probe fail | |
| + // (and terminate). Since the probe is in charge of cancelling all tracked | |
| + // requests, we must ensure that this probe sees our request. Adding the | |
| + // request prior to calling Signal() means that if we see an untripped | |
| + // breaker, no probe is running - consequently should the breaker then trip, | |
| + // it will observe our cancel, thus avoiding a leak. If we observe a tripped | |
| + // breaker, we also need to remove our own cancel, as the probe may already | |
| + // have passed the point at which it iterates through the cancels prior to us | |
| + // inserting it. The cancel may be invoked twice, but that's ok. | |
| + // | |
| + // See TestReplicaCircuitBreaker_NoCancelRace. | |
| + tok := br.cancels.Set(ctx, cancel) | |
| + if err := brSig.Err(); err != nil { | |
| + br.cancels.Del(tok) | |
| + cancel() | |
| + return nil, nil, err | |
| + } | |
| + | |
| + return tok, brSig, nil | |
| +} | |
| + | |
| +// UnregisterAndAdjustError releases a tracked cancel function upon request | |
| +// completion. The error resulting from the request is passed in to allow | |
| +// decorating it in case the breaker tripped while the request was in-flight. | |
| +// | |
| +// See Register. | |
| +func (br *replicaCircuitBreaker) UnregisterAndAdjustError( | |
| + tok interface{}, sig signaller, pErr *roachpb.Error, | |
| +) *roachpb.Error { | |
| + if sig.C() == nil { | |
| + // Breakers were disabled and we never put the cancel in the registry. | |
| + return pErr | |
| + } | |
| + | |
| + br.cancels.Del(tok) | |
| + | |
| + brErr := sig.Err() | |
| + if pErr == nil || brErr == nil { | |
| + return pErr | |
| + } | |
| + | |
| + // The breaker tripped and the command is returning an error. Make sure the | |
| + // error reflects the tripped breaker. | |
| + | |
| + err := pErr.GoError() | |
| + if ae := (&roachpb.AmbiguousResultError{}); errors.As(err, &ae) { | |
| + // The breaker tripped while a command was inflight, so we have to | |
| + // propagate an ambiguous result. We don't want to replace it, but there | |
| + // is a way to stash an Error in it so we use that. | |
| + // | |
| + // TODO(tbg): could also wrap it; there is no other write to WrappedErr | |
| + // in the codebase and it might be better to remove it. Nested *Errors | |
| + // are not a good idea. | |
| + wrappedErr := brErr | |
| + if ae.WrappedErr != nil { | |
| + wrappedErr = errors.Wrapf(brErr, "%v", ae.WrappedErr) | |
| + } | |
| + ae.WrappedErr = roachpb.NewError(wrappedErr) | |
| + return roachpb.NewError(ae) | |
| + } else if le := (&roachpb.NotLeaseHolderError{}); errors.As(err, &le) { | |
| + // When a lease acquisition triggered by this request is short-circuited | |
| + // by the breaker, it will return an opaque NotLeaseholderError, which we | |
| + // replace with the breaker's error. | |
| + return roachpb.NewError(errors.CombineErrors(brErr, le)) | |
| + } | |
| + return pErr | |
| +} | |
| + | |
| func (br *replicaCircuitBreaker) HasMark(err error) bool { | |
| return br.wrapped.HasMark(err) | |
| } | |
| +func (br *replicaCircuitBreaker) cancelAllTrackedContexts() { | |
| + br.cancels.Visit(func(ctx context.Context, cancel func()) (remove bool) { | |
| + cancel() | |
| + return true // remove | |
| + }) | |
| +} | |
| + | |
| func (br *replicaCircuitBreaker) canEnable() bool { | |
| b := atomic.LoadInt32(&br.versionIsActive) == 1 | |
| if b { | |
| @@ -102,7 +213,11 @@ func (br *replicaCircuitBreaker) enabled() bool { | |
| return replicaCircuitBreakerSlowReplicationThreshold.Get(&br.st.SV) > 0 && br.canEnable() | |
| } | |
| -func (br *replicaCircuitBreaker) TripAsync(err error) { | |
| +func (br *replicaCircuitBreaker) newError() error { | |
| + return br.r.replicaUnavailableError() | |
| +} | |
| + | |
| +func (br *replicaCircuitBreaker) TripAsync() { | |
| if !br.enabled() { | |
| return | |
| } | |
| @@ -110,15 +225,11 @@ func (br *replicaCircuitBreaker) TripAsync(err error) { | |
| _ = br.stopper.RunAsyncTask( | |
| br.ambCtx.AnnotateCtx(context.Background()), "trip-breaker", | |
| func(ctx context.Context) { | |
| - br.tripSync(err) | |
| + br.wrapped.Report(br.newError()) | |
| }, | |
| ) | |
| } | |
| -func (br *replicaCircuitBreaker) tripSync(err error) { | |
| - br.wrapped.Report(br.r.replicaUnavailableError(err)) | |
| -} | |
| - | |
| type signaller interface { | |
| Err() error | |
| C() <-chan struct{} | |
| @@ -141,6 +252,7 @@ func newReplicaCircuitBreaker( | |
| stopper *stop.Stopper, | |
| ambientCtx log.AmbientContext, | |
| r replicaInCircuitBreaker, | |
| + s CancelStorage, | |
| onTrip func(), | |
| onReset func(), | |
| ) *replicaCircuitBreaker { | |
| @@ -150,6 +262,8 @@ func newReplicaCircuitBreaker( | |
| r: r, | |
| st: cs, | |
| } | |
| + br.cancels = s | |
| + br.cancels.Reset() | |
| br.wrapped = circuit.NewBreaker(circuit.Options{ | |
| Name: "breaker", // log bridge has ctx tags | |
| AsyncProbe: br.asyncProbe, | |
| @@ -185,6 +299,16 @@ func (r replicaCircuitBreakerLogger) OnReset(br *circuit.Breaker) { | |
| r.EventHandler.OnReset(br) | |
| } | |
| +type probeKey struct{} | |
| + | |
| +func hasBypassCircuitBreakerMarker(ctx context.Context) bool { | |
| + return ctx.Value(probeKey{}) != nil | |
| +} | |
| + | |
| +func withBypassCircuitBreakerMarker(ctx context.Context) context.Context { | |
| + return context.WithValue(ctx, probeKey{}, probeKey{}) | |
| +} | |
| + | |
| func (br *replicaCircuitBreaker) asyncProbe(report func(error), done func()) { | |
| bgCtx := br.ambCtx.AnnotateCtx(context.Background()) | |
| if err := br.stopper.RunAsyncTask(bgCtx, "replica-probe", func(ctx context.Context) { | |
| @@ -195,19 +319,13 @@ func (br *replicaCircuitBreaker) asyncProbe(report func(error), done func()) { | |
| return | |
| } | |
| - brErr := br.Signal().Err() | |
| - if brErr == nil { | |
| - // This shouldn't happen, but if we're not even tripped, don't do | |
| - // anything. | |
| - return | |
| - } | |
| - | |
| - // Poison any inflight latches. Note that any new request that is added in | |
| - // while the probe is running but after poisonInflightLatches has been | |
| - // invoked will remain untouched. We rely on the replica to periodically | |
| - // access the circuit breaker to trigger additional probes in that case. | |
| - // (This happens in refreshProposalsLocked). | |
| - br.r.poisonInflightLatches(brErr) | |
| + // First, tell all current requests to fail fast. Note that clients insert | |
| + // first, then check the breaker (and remove themselves if breaker already | |
| + // tripped then). This prevents any cancels from sneaking in after the probe | |
| + // gets past this point, which could otherwise leave cancels hanging until | |
| + // "something" triggers the next probe (which may be never if no more traffic | |
| + // arrives at the Replica). See Register. | |
| + br.cancelAllTrackedContexts() | |
| err := sendProbe(ctx, br.r) | |
| report(err) | |
| }); err != nil { | |
| @@ -216,9 +334,10 @@ func (br *replicaCircuitBreaker) asyncProbe(report func(error), done func()) { | |
| } | |
| func sendProbe(ctx context.Context, r replicaInCircuitBreaker) error { | |
| - // NB: ProbeRequest has the bypassesCircuitBreaker flag. If in the future we | |
| - // enhance the probe, we may need to allow any additional requests we send to | |
| - // chose to bypass the circuit breaker explicitly. | |
| + // NB: we don't need to put this marker since ProbeRequest has the | |
| + // canBypassReplicaCircuitBreaker flag, but if in the future we do | |
| + // additional work in this method we may need it. | |
| + ctx = withBypassCircuitBreakerMarker(ctx) | |
| desc := r.Desc() | |
| if !desc.IsInitialized() { | |
| return nil | |
| @@ -240,13 +359,12 @@ func sendProbe(ctx context.Context, r replicaInCircuitBreaker) error { | |
| return pErr.GoError() | |
| }, | |
| ); err != nil { | |
| - return r.replicaUnavailableError(err) | |
| + return errors.CombineErrors(r.replicaUnavailableError(), err) | |
| } | |
| return nil | |
| } | |
| func replicaUnavailableError( | |
| - err error, | |
| desc *roachpb.RangeDescriptor, | |
| replDesc roachpb.ReplicaDescriptor, | |
| lm liveness.IsLiveMap, | |
| @@ -271,22 +389,25 @@ func replicaUnavailableError( | |
| var _ redact.SafeFormatter = desc | |
| var _ redact.SafeFormatter = replDesc | |
| - if len(nonLiveRepls.AsProto()) > 0 { | |
| - err = errors.Wrapf(err, "replicas on non-live nodes: %v (lost quorum: %t)", nonLiveRepls, !canMakeProgress) | |
| - } | |
| - | |
| + err := roachpb.NewReplicaUnavailableError(desc, replDesc) | |
| err = errors.Wrapf( | |
| err, | |
| "raft status: %+v", redact.Safe(rs), // raft status contains no PII | |
| ) | |
| + if len(nonLiveRepls.AsProto()) > 0 { | |
| + err = errors.Wrapf(err, "replicas on non-live nodes: %v (lost quorum: %t)", nonLiveRepls, !canMakeProgress) | |
| + } | |
| - return roachpb.NewReplicaUnavailableError(err, desc, replDesc) | |
| + return err | |
| } | |
| -func (r *Replica) replicaUnavailableError(err error) error { | |
| +func (r *Replica) replicaUnavailableError() error { | |
| desc := r.Desc() | |
| replDesc, _ := desc.GetReplicaDescriptor(r.store.StoreID()) | |
| - isLiveMap, _ := r.store.livenessMap.Load().(liveness.IsLiveMap) | |
| - return replicaUnavailableError(err, desc, replDesc, isLiveMap, r.RaftStatus()) | |
| + var isLiveMap liveness.IsLiveMap | |
| + if nl := r.store.cfg.NodeLiveness; nl != nil { // exclude unit test | |
| + isLiveMap = nl.GetIsLiveMap() | |
| + } | |
| + return replicaUnavailableError(desc, replDesc, isLiveMap, r.RaftStatus()) | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_circuit_breaker_cancelstorage.go b/pkg/kv/kvserver/replica_circuit_breaker_cancelstorage.go | |
| new file mode 100644 | |
| index 0000000000..c8f46bcffa | |
| --- /dev/null | |
| +++ b/pkg/kv/kvserver/replica_circuit_breaker_cancelstorage.go | |
| @@ -0,0 +1,115 @@ | |
| +// Copyright 2022 The Cockroach Authors. | |
| +// | |
| +// Use of this software is governed by the Business Source License | |
| +// included in the file licenses/BSL.txt. | |
| +// | |
| +// As of the Change Date specified in that file, in accordance with | |
| +// the Business Source License, use of this software will be governed | |
| +// by the Apache License, Version 2.0, included in the file | |
| +// licenses/APL.txt. | |
| + | |
| +package kvserver | |
| + | |
| +import ( | |
| + "context" | |
| + "sync" | |
| + "unsafe" | |
| + | |
| + "github.com/cockroachdb/cockroach/pkg/util/syncutil" | |
| +) | |
| + | |
| +// CancelStorage implements tracking of context cancellation functions | |
| +// for use by Replica circuit breakers. | |
| +type CancelStorage interface { | |
| + // Reset initializes the storage. Not thread safe. | |
| + Reset() | |
| + // Set adds context and associated cancel func to the storage. Returns a token | |
| + // that can be passed to Del. | |
| + // | |
| + // Set is thread-safe. | |
| + Set(_ context.Context, cancel func()) (token interface{}) | |
| + // Del removes a cancel func, as identified by the token returned from Set. | |
| + // | |
| + // Del is thread-safe. | |
| + Del(token interface{}) | |
| + // Visit invokes the provided closure with each (context,cancel) pair currently | |
| + // present in the storage. Items for which the visitor returns true are removed | |
| + // from the storage. | |
| + // | |
| + // Visit is thread-safe, but it is illegal to invoke methods of the | |
| + // CancelStorage from within the visitor. | |
| + Visit(func(context.Context, func()) (remove bool)) | |
| +} | |
| + | |
| +type cancelToken struct { | |
| + ctx context.Context | |
| +} | |
| + | |
| +func (tok *cancelToken) fasthash() int { | |
| + // From https://github.com/taylorza/go-lfsr/blob/7ec2b93980f950da1e36c6682771e6fe14c144c2/lfsr.go#L46-L48. | |
| + s := int(uintptr(unsafe.Pointer(tok))) | |
| + b := (s >> 0) ^ (s >> 2) ^ (s >> 3) ^ (s >> 4) | |
| + return (s >> 1) | (b << 7) | |
| +} | |
| + | |
| +var cancelTokenPool = sync.Pool{ | |
| + New: func() interface{} { return &cancelToken{} }, | |
| +} | |
| + | |
| +type mapCancelShard struct { | |
| + syncutil.Mutex | |
| + m map[*cancelToken]func() | |
| +} | |
| + | |
| +// A MapCancelStorage implements CancelStorage via shards of mutex-protected | |
| +// maps. | |
| +type MapCancelStorage struct { | |
| + NumShards int | |
| + sl []*mapCancelShard | |
| +} | |
| + | |
| +// Reset implements CancelStorage. | |
| +func (m *MapCancelStorage) Reset() { | |
| + if m.NumShards == 0 { | |
| + m.NumShards = 1 | |
| + } | |
| + m.sl = make([]*mapCancelShard, m.NumShards) | |
| + for i := range m.sl { | |
| + s := &mapCancelShard{} | |
| + s.m = map[*cancelToken]func(){} | |
| + m.sl[i] = s | |
| + } | |
| +} | |
| + | |
| +// Set implements CancelStorage. | |
| +func (m *MapCancelStorage) Set(ctx context.Context, cancel func()) interface{} { | |
| + tok := cancelTokenPool.Get().(*cancelToken) | |
| + tok.ctx = ctx | |
| + shard := m.sl[tok.fasthash()%len(m.sl)] | |
| + shard.Lock() | |
| + shard.m[tok] = cancel | |
| + shard.Unlock() | |
| + return tok | |
| +} | |
| + | |
| +// Del implements CancelStorage. | |
| +func (m *MapCancelStorage) Del(tok interface{}) { | |
| + ttok := tok.(*cancelToken) | |
| + shard := m.sl[ttok.fasthash()%len(m.sl)] | |
| + shard.Lock() | |
| + delete(shard.m, tok.(*cancelToken)) | |
| + shard.Unlock() | |
| +} | |
| + | |
| +// Visit implements CancelStorage. | |
| +func (m *MapCancelStorage) Visit(fn func(context.Context, func()) (remove bool)) { | |
| + for _, shard := range m.sl { | |
| + shard.Lock() | |
| + for tok, cancel := range shard.m { | |
| + if fn(tok.ctx, cancel) { | |
| + delete(shard.m, tok) | |
| + } | |
| + } | |
| + shard.Unlock() | |
| + } | |
| +} | |
| diff --git a/pkg/kv/kvserver/replica_circuit_breaker_test.go b/pkg/kv/kvserver/replica_circuit_breaker_test.go | |
| index ab1f9aa32d..c7b6029b7d 100644 | |
| --- a/pkg/kv/kvserver/replica_circuit_breaker_test.go | |
| +++ b/pkg/kv/kvserver/replica_circuit_breaker_test.go | |
| @@ -12,14 +12,23 @@ package kvserver | |
| import ( | |
| "context" | |
| + "fmt" | |
| + "math/rand" | |
| + "runtime" | |
| + "strconv" | |
| "testing" | |
| + "time" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/liveness" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| + "github.com/cockroachdb/cockroach/pkg/settings/cluster" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/echotest" | |
| + "github.com/cockroachdb/cockroach/pkg/util/ctxgroup" | |
| + "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/cockroach/pkg/util/leaktest" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| + "github.com/cockroachdb/cockroach/pkg/util/stop" | |
| "github.com/cockroachdb/errors" | |
| "github.com/cockroachdb/redact" | |
| "github.com/stretchr/testify/require" | |
| @@ -39,12 +48,155 @@ func TestReplicaUnavailableError(t *testing.T) { | |
| lm := liveness.IsLiveMap{ | |
| 1: liveness.IsLiveMapEntry{IsLive: true}, | |
| } | |
| - wrappedErr := errors.New("probe failed") | |
| rs := raft.Status{} | |
| - ctx := context.Background() | |
| - err := errors.DecodeError(ctx, errors.EncodeError(ctx, replicaUnavailableError( | |
| - wrappedErr, desc, desc.Replicas().AsProto()[0], lm, &rs), | |
| - )) | |
| - require.True(t, errors.Is(err, wrappedErr), "%+v", err) | |
| + err := replicaUnavailableError(desc, desc.Replicas().AsProto()[0], lm, &rs) | |
| echotest.Require(t, string(redact.Sprint(err)), testutils.TestDataPath(t, "replica_unavailable_error.txt")) | |
| } | |
| + | |
| +type circuitBreakerReplicaMock struct { | |
| + clock *hlc.Clock | |
| +} | |
| + | |
| +func (c *circuitBreakerReplicaMock) Clock() *hlc.Clock { | |
| + return c.clock | |
| +} | |
| + | |
| +func (c *circuitBreakerReplicaMock) Desc() *roachpb.RangeDescriptor { | |
| + return &roachpb.RangeDescriptor{} | |
| +} | |
| + | |
| +func (c *circuitBreakerReplicaMock) Send( | |
| + ctx context.Context, ba roachpb.BatchRequest, | |
| +) (*roachpb.BatchResponse, *roachpb.Error) { | |
| + return ba.CreateReply(), nil | |
| +} | |
| + | |
| +func (c *circuitBreakerReplicaMock) slowReplicationThreshold( | |
| + ba *roachpb.BatchRequest, | |
| +) (time.Duration, bool) { | |
| + return 0, false | |
| +} | |
| + | |
| +func (c *circuitBreakerReplicaMock) replicaUnavailableError() error { | |
| + return errors.New("unavailable") | |
| +} | |
| + | |
| +// This test verifies that when the breaker trips and untrips again, | |
| +// there is no scenario under which the request's cancel leaks. | |
| +func TestReplicaCircuitBreaker_NoCancelRace(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + br, stopper := setupCircuitBreakerTest(t, "mutexmap-1") | |
| + ctx, cancel := context.WithCancel(context.Background()) | |
| + defer cancel() | |
| + defer stopper.Stop(ctx) | |
| + | |
| + g := ctxgroup.WithContext(ctx) | |
| + const count = 100 | |
| + for i := 0; i < count; i++ { | |
| + i := i // for goroutine | |
| + g.GoCtx(func(ctx context.Context) error { | |
| + ctx, cancel := context.WithCancel(ctx) | |
| + tok, sig, err := br.Register(ctx, cancel) | |
| + if err != nil { | |
| + _ = err // ignoring intentionally | |
| + return nil | |
| + } | |
| + if i == count/2 { | |
| + br.TripAsync() // probe will succeed | |
| + } | |
| + runtime.Gosched() | |
| + time.Sleep(time.Duration(rand.Intn(int(time.Millisecond)))) | |
| + var pErr *roachpb.Error | |
| + if i%2 == 0 { | |
| + pErr = roachpb.NewErrorf("boom") | |
| + } | |
| + _ = br.UnregisterAndAdjustError(tok, sig, pErr) | |
| + return nil | |
| + }) | |
| + } | |
| + require.NoError(t, g.Wait()) | |
| + var n int | |
| + br.cancels.Visit(func(ctx context.Context, _ func()) (remove bool) { | |
| + n++ | |
| + return false // keep | |
| + }) | |
| + require.Zero(t, n, "found tracked requests") | |
| +} | |
| + | |
| +func TestReplicaCircuitBreaker_Register(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + br, stopper := setupCircuitBreakerTest(t, "mutexmap-1") | |
| + defer stopper.Stop(context.Background()) | |
| + ctx := withBypassCircuitBreakerMarker(context.Background()) | |
| + tok, sig, err := br.Register(ctx, func() {}) | |
| + require.NoError(t, err) | |
| + defer br.UnregisterAndAdjustError(tok, sig, nil /* pErr */) | |
| + require.Zero(t, sig.C()) | |
| + var n int | |
| + br.cancels.Visit(func(ctx context.Context, f func()) (remove bool) { | |
| + n++ | |
| + return false // keep | |
| + }) | |
| + require.Zero(t, n, "probe context got added to CancelStorage") | |
| +} | |
| + | |
| +func setupCircuitBreakerTest(t testing.TB, cs string) (*replicaCircuitBreaker, *stop.Stopper) { | |
| + st := cluster.MakeTestingClusterSettings() | |
| + // Enable circuit breakers. | |
| + replicaCircuitBreakerSlowReplicationThreshold.Override(context.Background(), &st.SV, time.Hour) | |
| + r := &circuitBreakerReplicaMock{clock: hlc.NewClock(hlc.UnixNano, 500*time.Millisecond)} | |
| + var numShards int | |
| + { | |
| + _, err := fmt.Sscanf(cs, "mutexmap-%d", &numShards) | |
| + require.NoError(t, err) | |
| + } | |
| + s := &MapCancelStorage{NumShards: numShards} | |
| + onTrip := func() {} | |
| + onReset := func() {} | |
| + stopper := stop.NewStopper() | |
| + br := newReplicaCircuitBreaker(st, stopper, log.AmbientContext{}, r, s, onTrip, onReset) | |
| + return br, stopper | |
| +} | |
| + | |
| +func BenchmarkReplicaCircuitBreaker_Register(b *testing.B) { | |
| + defer leaktest.AfterTest(b)() | |
| + | |
| + for _, enabled := range []bool{false, true} { | |
| + b.Run("enabled="+strconv.FormatBool(enabled), func(b *testing.B) { | |
| + dss := []string{ | |
| + "mutexmap-1", "mutexmap-2", "mutexmap-4", "mutexmap-8", "mutexmap-12", "mutexmap-16", | |
| + "mutexmap-20", "mutexmap-24", "mutexmap-32", "mutexmap-64", | |
| + } | |
| + if !enabled { | |
| + dss = dss[:1] | |
| + } | |
| + for _, ds := range dss { | |
| + b.Run(ds, func(b *testing.B) { | |
| + b.ReportAllocs() | |
| + br, stopper := setupCircuitBreakerTest(b, ds) | |
| + defer stopper.Stop(context.Background()) | |
| + | |
| + var dur time.Duration | |
| + if enabled { | |
| + dur = time.Hour | |
| + } | |
| + replicaCircuitBreakerSlowReplicationThreshold.Override(context.Background(), &br.st.SV, dur) | |
| + b.ResetTimer() | |
| + b.RunParallel(func(pb *testing.PB) { | |
| + for pb.Next() { | |
| + ctx, cancel := context.WithCancel(context.Background()) | |
| + tok, sig, err := br.Register(ctx, cancel) | |
| + if err != nil { | |
| + b.Error(err) | |
| + } | |
| + if pErr := br.UnregisterAndAdjustError(tok, sig, nil); pErr != nil { | |
| + b.Error(pErr) | |
| + } | |
| + cancel() | |
| + } | |
| + }) | |
| + }) | |
| + } | |
| + }) | |
| + } | |
| +} | |
| diff --git a/pkg/kv/kvserver/replica_closedts_test.go b/pkg/kv/kvserver/replica_closedts_test.go | |
| index 97692e703b..6726105910 100644 | |
| --- a/pkg/kv/kvserver/replica_closedts_test.go | |
| +++ b/pkg/kv/kvserver/replica_closedts_test.go | |
| @@ -28,7 +28,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/server" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/serverutils" | |
| - "github.com/cockroachdb/cockroach/pkg/testutils/skip" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/sqlutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/testcluster" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| @@ -46,9 +45,6 @@ import ( | |
| func TestBumpSideTransportClosed(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| - | |
| - skip.UnderRace(t) | |
| - | |
| ctx := context.Background() | |
| type setupArgs struct { | |
| diff --git a/pkg/kv/kvserver/replica_command.go b/pkg/kv/kvserver/replica_command.go | |
| index 70588bfa58..6ed71ebceb 100644 | |
| --- a/pkg/kv/kvserver/replica_command.go | |
| +++ b/pkg/kv/kvserver/replica_command.go | |
| @@ -3353,29 +3353,39 @@ func (r *Replica) adminScatter( | |
| } | |
| ri := r.GetRangeInfo(ctx) | |
| - stats := r.GetMVCCStats() | |
| return roachpb.AdminScatterResponse{ | |
| RangeInfos: []roachpb.RangeInfo{ri}, | |
| - MVCCStats: &stats, | |
| }, nil | |
| } | |
| -// TODO(arul): AdminVerifyProtectedTimestampRequest can entirely go away in | |
| -// 22.2. | |
| func (r *Replica) adminVerifyProtectedTimestamp( | |
| - ctx context.Context, _ roachpb.AdminVerifyProtectedTimestampRequest, | |
| + ctx context.Context, args roachpb.AdminVerifyProtectedTimestampRequest, | |
| ) (resp roachpb.AdminVerifyProtectedTimestampResponse, err error) { | |
| - // AdminVerifyProtectedTimestampRequest is not supported starting from the | |
| - // 22.1 release. We expect nodes running a 22.1 binary to still service this | |
| - // request in a {21.2, 22.1} mixed version cluster. This can happen if the | |
| - // request is initiated on a 21.2 node and the leaseholder of the range it is | |
| - // trying to verify is on a 22.1 node. | |
| - // | |
| - // We simply return true without attempting to verify in such a case. This | |
| - // ensures upstream jobs (backups) don't fail as a result. It is okay to | |
| - // return true regardless even if the PTS record being verified does not apply | |
| - // as the failure mode is non-destructive. Infact, this is the reason we're | |
| - // no longer supporting Verification past 22.1. | |
| - resp.Verified = true | |
| + var doesNotApplyReason string | |
| + resp.Verified, doesNotApplyReason, err = r.protectedTimestampRecordApplies(ctx, &args) | |
| + if err != nil { | |
| + return resp, err | |
| + } | |
| + | |
| + // In certain cases we do not want to return an error even if we failed to | |
| + // verify the protected ts record. This ensures that executeAdminBatch adds | |
| + // the response to the batch, thereby allowing us to aggregate the | |
| + // verification failures across all AdminVerifyProtectedTimestampRequests and | |
| + // construct a more informative error to show to the user. | |
| + if doesNotApplyReason != "" { | |
| + if !resp.Verified { | |
| + desc := r.Desc() | |
| + failedRange := roachpb.AdminVerifyProtectedTimestampResponse_FailedRange{ | |
| + RangeID: int64(desc.GetRangeID()), | |
| + StartKey: desc.GetStartKey(), | |
| + EndKey: desc.EndKey, | |
| + Reason: doesNotApplyReason, | |
| + } | |
| + resp.VerificationFailedRanges = append(resp.VerificationFailedRanges, failedRange) | |
| + // TODO(adityamaru): This is here for compatibility with 20.2, remove in | |
| + // 21.2. | |
| + resp.DeprecatedFailedRanges = append(resp.DeprecatedFailedRanges, *r.Desc()) | |
| + } | |
| + } | |
| return resp, nil | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_consistency.go b/pkg/kv/kvserver/replica_consistency.go | |
| index fd270fe5ef..3fc943c631 100644 | |
| --- a/pkg/kv/kvserver/replica_consistency.go | |
| +++ b/pkg/kv/kvserver/replica_consistency.go | |
| @@ -253,7 +253,7 @@ func (r *Replica) CheckConsistency( | |
| if !haveDelta { | |
| return resp, nil | |
| } | |
| - log.Fatalf(ctx, "found a delta of %+v", redact.Safe(delta)) | |
| + log.Fatalf(ctx, "found a delta of %+v", log.Safe(delta)) | |
| } | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_destroy.go b/pkg/kv/kvserver/replica_destroy.go | |
| index a6a8ed9fd5..cdc32adb3e 100644 | |
| --- a/pkg/kv/kvserver/replica_destroy.go | |
| +++ b/pkg/kv/kvserver/replica_destroy.go | |
| @@ -209,7 +209,8 @@ func (r *Replica) disconnectReplicationRaftMuLocked(ctx context.Context) { | |
| // share the error across proposals). | |
| p.finishApplication(ctx, proposalResult{ | |
| Err: roachpb.NewError( | |
| - roachpb.NewAmbiguousResultError(apply.ErrRemoved)), | |
| + roachpb.NewAmbiguousResultError( | |
| + apply.ErrRemoved.Error())), | |
| }) | |
| } | |
| r.mu.internalRaftGroup = nil | |
| diff --git a/pkg/kv/kvserver/replica_init.go b/pkg/kv/kvserver/replica_init.go | |
| index 5f7e1629d3..2545b21644 100644 | |
| --- a/pkg/kv/kvserver/replica_init.go | |
| +++ b/pkg/kv/kvserver/replica_init.go | |
| @@ -137,8 +137,14 @@ func newUnloadedReplica( | |
| onReset := func() { | |
| store.Metrics().ReplicaCircuitBreakerCurTripped.Dec(1) | |
| } | |
| + var cancelStorage CancelStorage | |
| + if f := r.store.cfg.TestingKnobs.CancelStorageFactory; f != nil { | |
| + cancelStorage = f() | |
| + } else { | |
| + cancelStorage = &MapCancelStorage{} | |
| + } | |
| r.breaker = newReplicaCircuitBreaker( | |
| - store.cfg.Settings, store.stopper, r.AmbientContext, r, onTrip, onReset, | |
| + store.cfg.Settings, store.stopper, r.AmbientContext, r, cancelStorage, onTrip, onReset, | |
| ) | |
| return r | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_learner_test.go b/pkg/kv/kvserver/replica_learner_test.go | |
| index 3edeb38dae..9599fae010 100644 | |
| --- a/pkg/kv/kvserver/replica_learner_test.go | |
| +++ b/pkg/kv/kvserver/replica_learner_test.go | |
| @@ -27,7 +27,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverpb" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/server" | |
| - "github.com/cockroachdb/cockroach/pkg/server/serverpb" | |
| "github.com/cockroachdb/cockroach/pkg/testutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/serverutils" | |
| "github.com/cockroachdb/cockroach/pkg/testutils/skip" | |
| @@ -313,7 +312,6 @@ func TestLearnerSnapshotFailsRollback(t *testing.T) { | |
| defer log.Scope(t).Close(t) | |
| skip.UnderShort(t) // Takes 90s. | |
| - skip.UnderRace(t) | |
| runTest := func(t *testing.T, replicaType roachpb.ReplicaType) { | |
| var rejectSnapshots int64 | |
| @@ -360,7 +358,15 @@ func TestLearnerSnapshotFailsRollback(t *testing.T) { | |
| }) | |
| } | |
| -func testRaftSnapshotsToNonVoters(t *testing.T, drainReceivingNode bool) { | |
| +// TestNonVoterCatchesUpViaRaftSnapshotQueue ensures that a non-voting replica | |
| +// in need of a snapshot will receive one via the raft snapshot queue. This is | |
| +// also meant to test that a non-voting replica that is initialized via an | |
| +// `INITIAL` snapshot during its addition is not ignored by the raft snapshot | |
| +// queue for future snapshots. | |
| +func TestNonVoterCatchesUpViaRaftSnapshotQueue(t *testing.T) { | |
| + defer leaktest.AfterTest(t)() | |
| + defer log.Scope(t).Close(t) | |
| + | |
| skip.UnderShort(t, "this test sleeps for a few seconds") | |
| var skipInitialSnapshot int64 | |
| @@ -381,7 +387,6 @@ func testRaftSnapshotsToNonVoters(t *testing.T, drainReceivingNode bool) { | |
| // Disable the raft snapshot queue, we will manually queue a replica into it | |
| // below. | |
| ltk.storeKnobs.DisableRaftSnapshotQueue = true | |
| - | |
| tc := testcluster.StartTestCluster( | |
| t, 2, base.TestClusterArgs{ | |
| ServerArgs: base.TestServerArgs{Knobs: knobs}, | |
| @@ -401,8 +406,6 @@ func testRaftSnapshotsToNonVoters(t *testing.T, drainReceivingNode bool) { | |
| return err | |
| }) | |
| - // Wait until we remove the lock that prevents the raft snapshot queue from | |
| - // sending this replica a snapshot. | |
| select { | |
| case <-nonVoterSnapLockRemoved: | |
| case <-time.After(testutils.DefaultSucceedsSoonDuration): | |
| @@ -418,16 +421,6 @@ func testRaftSnapshotsToNonVoters(t *testing.T, drainReceivingNode bool) { | |
| time.Sleep(kvserver.RaftLogQueuePendingSnapshotGracePeriod) | |
| - if drainReceivingNode { | |
| - // Draining nodes shouldn't reject raft snapshots, so this should have no | |
| - // effect on the outcome of this test. | |
| - const drainingServerIdx = 1 | |
| - const drainingNodeID = drainingServerIdx + 1 | |
| - client, err := tc.GetAdminClient(ctx, t, drainingServerIdx) | |
| - require.NoError(t, err) | |
| - drain(ctx, t, client, drainingNodeID) | |
| - } | |
| - | |
| testutils.SucceedsSoon(t, func() error { | |
| // Manually enqueue the leaseholder replica into its store's raft snapshot | |
| // queue. We expect it to pick up on the fact that the non-voter on its range | |
| @@ -453,62 +446,6 @@ func testRaftSnapshotsToNonVoters(t *testing.T, drainReceivingNode bool) { | |
| require.NoError(t, g.Wait()) | |
| } | |
| -func drain(ctx context.Context, t *testing.T, client serverpb.AdminClient, drainingNodeID int) { | |
| - stream, err := client.Drain(ctx, &serverpb.DrainRequest{ | |
| - NodeId: strconv.Itoa(drainingNodeID), | |
| - DoDrain: true, | |
| - }) | |
| - require.NoError(t, err) | |
| - | |
| - // Wait until the draining node acknowledges that it's draining. | |
| - _, err = stream.Recv() | |
| - require.NoError(t, err) | |
| -} | |
| - | |
| -// TestSnapshotsToDrainingNodes tests that rebalancing snapshots to draining | |
| -// receivers are rejected, but Raft snapshots aren't. | |
| -func TestSnapshotsToDrainingNodes(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - | |
| - t.Run("rebalancing snapshots", func(t *testing.T) { | |
| - ctx := context.Background() | |
| - | |
| - // We set up a 2 node test cluster with the second node marked draining. | |
| - const drainingServerIdx = 1 | |
| - const drainingNodeID = drainingServerIdx + 1 | |
| - tc := testcluster.StartTestCluster( | |
| - t, 2, base.TestClusterArgs{ | |
| - ReplicationMode: base.ReplicationManual, | |
| - }, | |
| - ) | |
| - defer tc.Stopper().Stop(ctx) | |
| - client, err := tc.GetAdminClient(ctx, t, drainingServerIdx) | |
| - require.NoError(t, err) | |
| - drain(ctx, t, client, drainingNodeID) | |
| - | |
| - // Now, we try to add a replica to it, we expect that to fail. | |
| - scratchKey := tc.ScratchRange(t) | |
| - _, err = tc.AddVoters(scratchKey, makeReplicationTargets(drainingNodeID)...) | |
| - require.Regexp(t, "store is draining", err) | |
| - }) | |
| - | |
| - t.Run("raft snapshots", func(t *testing.T) { | |
| - testRaftSnapshotsToNonVoters(t, true /* drainReceivingNode */) | |
| - }) | |
| -} | |
| - | |
| -// TestNonVoterCatchesUpViaRaftSnapshotQueue ensures that a non-voting replica | |
| -// in need of a snapshot will receive one via the raft snapshot queue. This is | |
| -// also meant to test that a non-voting replica that is initialized via an | |
| -// `INITIAL` snapshot during its addition is not ignored by the raft snapshot | |
| -// queue for future snapshots. | |
| -func TestNonVoterCatchesUpViaRaftSnapshotQueue(t *testing.T) { | |
| - defer leaktest.AfterTest(t)() | |
| - defer log.Scope(t).Close(t) | |
| - testRaftSnapshotsToNonVoters(t, false /* drainReceivingNode */) | |
| -} | |
| - | |
| func TestSplitWithLearnerOrJointConfig(t *testing.T) { | |
| defer leaktest.AfterTest(t)() | |
| defer log.Scope(t).Close(t) | |
| diff --git a/pkg/kv/kvserver/replica_proposal.go b/pkg/kv/kvserver/replica_proposal.go | |
| index fa08b6fc22..bbef1c9cfa 100644 | |
| --- a/pkg/kv/kvserver/replica_proposal.go | |
| +++ b/pkg/kv/kvserver/replica_proposal.go | |
| @@ -35,7 +35,6 @@ import ( | |
| "github.com/cockroachdb/cockroach/pkg/util/quotapool" | |
| "github.com/cockroachdb/cockroach/pkg/util/timeutil" | |
| "github.com/cockroachdb/cockroach/pkg/util/tracing" | |
| - "github.com/cockroachdb/redact" | |
| "github.com/kr/pretty" | |
| "golang.org/x/time/rate" | |
| ) | |
| @@ -159,16 +158,6 @@ func (proposal *ProposalData) signalProposalResult(pr proposalResult) { | |
| if proposal.doneCh != nil { | |
| proposal.doneCh <- pr | |
| proposal.doneCh = nil | |
| - // Need to remove any span from the proposal, as the signalled caller | |
| - // will likely finish it, and if we then end up applying this proposal | |
| - // we'll try to make a ChildSpan off `proposal.ctx` and this will | |
| - // trigger the Span use-after-finish assertions. | |
| - // | |
| - // See: https://github.com/cockroachdb/cockroach/pull/76858#issuecomment-1048179588 | |
| - // | |
| - // NB: `proposal.ec.repl` might already have been cleared if we arrive here | |
| - // through finishApplication. | |
| - proposal.ctx = context.Background() | |
| } | |
| } | |
| @@ -237,7 +226,7 @@ func (r *Replica) leasePostApplyLocked( | |
| switch { | |
| case s2 < s1: | |
| log.Fatalf(ctx, "lease sequence inversion, prevLease=%s, newLease=%s", | |
| - redact.Safe(prevLease), redact.Safe(newLease)) | |
| + log.Safe(prevLease), log.Safe(newLease)) | |
| case s2 == s1: | |
| // If the sequence numbers are the same, make sure they're actually | |
| // the same lease. This can happen when callers are using | |
| @@ -245,13 +234,13 @@ func (r *Replica) leasePostApplyLocked( | |
| // splitPostApply. It can also happen during lease extensions. | |
| if !prevLease.Equivalent(*newLease) { | |
| log.Fatalf(ctx, "sequence identical for different leases, prevLease=%s, newLease=%s", | |
| - redact.Safe(prevLease), redact.Safe(newLease)) | |
| + log.Safe(prevLease), log.Safe(newLease)) | |
| } | |
| case s2 == s1+1: | |
| // Lease sequence incremented by 1. Expected case. | |
| case s2 > s1+1 && jumpOpt == assertNoLeaseJump: | |
| log.Fatalf(ctx, "lease sequence jump, prevLease=%s, newLease=%s", | |
| - redact.Safe(prevLease), redact.Safe(newLease)) | |
| + log.Safe(prevLease), log.Safe(newLease)) | |
| } | |
| } | |
| diff --git a/pkg/kv/kvserver/replica_protected_timestamp.go b/pkg/kv/kvserver/replica_protected_timestamp.go | |
| index fb86e67198..cc23a79c78 100644 | |
| --- a/pkg/kv/kvserver/replica_protected_timestamp.go | |
| +++ b/pkg/kv/kvserver/replica_protected_timestamp.go | |
| @@ -12,9 +12,12 @@ package kvserver | |
| import ( | |
| "context" | |
| + "fmt" | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/kv/kvserver/gc" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/kvserverbase" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/protectedts/ptpb" | |
| "github.com/cockroachdb/cockroach/pkg/roachpb" | |
| "github.com/cockroachdb/cockroach/pkg/util/hlc" | |
| "github.com/cockroachdb/cockroach/pkg/util/log" | |
| @@ -24,8 +27,8 @@ import ( | |
| // cachedProtectedTimestampState is used to cache information about the state | |
| // of protected timestamps as they pertain to this replica. The data is | |
| // refreshed when the replica examines protected timestamps when being | |
| -// considered for gc. It is consulted when determining whether a request can be | |
| -// served. | |
| +// considered for gc or when verifying a protected timestamp record. | |
| +// It is consulted when determining whether a request can be served. | |
| type cachedProtectedTimestampState struct { | |
| // readAt denotes the timestamp at which this record was read. | |
| // It is used to coordinate updates to this field. It is also used to | |
| @@ -34,8 +37,8 @@ type cachedProtectedTimestampState struct { | |
| // that protected timestamps have not been observed. In this case we must | |
| // assume that any protected timestamp could exist to provide the contract | |
| // on verify. | |
| - readAt hlc.Timestamp | |
| - earliestProtectionTimestamp hlc.Timestamp | |
| + readAt hlc.Timestamp | |
| + earliestRecord *ptpb.Record | |
| } | |
| // clearIfNotNewer clears the state in ts if it is not newer than the passed | |
| @@ -64,35 +67,185 @@ func (r *Replica) maybeUpdateCachedProtectedTS(ts *cachedProtectedTimestampState | |
| } | |
| } | |
| +// protectedTimestampRecordApplies returns true if it is this case that the | |
| +// record which protects the `protected` timestamp. It returns false if it may | |
| +// not. If the state of the cache is not sufficiently new to determine whether | |
| +// the record will apply, the cache is refreshed and then the check is performed | |
| +// again. See r.protectedTimestampRecordCurrentlyApplies() for more details. | |
| +func (r *Replica) protectedTimestampRecordApplies( | |
| + ctx context.Context, args *roachpb.AdminVerifyProtectedTimestampRequest, | |
| +) (willApply bool, doesNotApplyReason string, _ error) { | |
| + // Check the state of the cache without a refresh. | |
| + willApply, cacheTooOld, doesNotApplyReason, err := r.protectedTimestampRecordCurrentlyApplies( | |
| + ctx, args) | |
| + if err != nil { | |
| + return false, doesNotApplyReason, err | |
| + } | |
| + if !cacheTooOld { | |
| + return willApply, doesNotApplyReason, nil | |
| + } | |
| + // Refresh the cache so that we know that the next time we come around we're | |
| + // certain to either see the record or see a timestamp for readAt that is | |
| + // greater than or equal to recordAliveAt. | |
| + if err := r.store.protectedtsCache.Refresh(ctx, args.RecordAliveAt); err != nil { | |
| + return false, doesNotApplyReason, err | |
| + } | |
| + willApply, cacheTooOld, doesNotApplyReason, err = r.protectedTimestampRecordCurrentlyApplies( | |
| + ctx, args) | |
| + if err != nil { | |
| + return false, doesNotApplyReason, err | |
| + } | |
| + if cacheTooOld { | |
| + return false, doesNotApplyReason, errors.AssertionFailedf( | |
| + "cache was not updated after being refreshed") | |
| + } | |
| + return willApply, doesNotApplyReason, nil | |
| +} | |
| + | |
| func (r *Replica) readProtectedTimestampsRLocked( | |
| - ctx context.Context, | |
| -) (ts cachedProtectedTimestampState, _ error) { | |
| + ctx context.Context, f func(r *ptpb.Record), | |
| +) (ts cachedProtectedTimestampState) { | |
| desc := r.descRLocked() | |
| gcThreshold := *r.mu.state.GCThreshold | |
| - sp := roachpb.Span{ | |
| - Key: roachpb.Key(desc.StartKey), | |
| - EndKey: roachpb.Key(desc.EndKey), | |
| + ts.readAt = r.store.protectedtsCache.Iterate(ctx, | |
| + roachpb.Key(desc.StartKey), | |
| + roachpb.Key(desc.EndKey), | |
| + func(rec *ptpb.Record) (wantMore bool) { | |
| + // Check if we've already GC'd past the timestamp this record was trying | |
| + // to protect, in which case we know that the record does not apply. | |
| + // Note that when we implement PROTECT_AT, we'll need to consult some | |
| + // replica state here to determine whether the record indeed has been | |
| + // applied. | |
| + if isValid := gcThreshold.LessEq(rec.Timestamp); !isValid { | |
| + return true | |
| + } | |
| + if f != nil { | |
| + f(rec) | |
| + } | |
| + if ts.earliestRecord == nil || rec.Timestamp.Less(ts.earliestRecord.Timestamp) { | |
| + ts.earliestRecord = rec | |
| + } | |
| + return true | |
| + }) | |
| + return ts | |
| +} | |
| + | |
| +// protectedTimestampRecordCurrentlyApplies determines whether a record with | |
| +// the specified ID which protects `protected` and is known to exist at | |
| +// `recordAliveAt` will apply given the current state of the cache. This method | |
| +// is called by `r.protectedTimestampRecordApplies()`. It may be the case that | |
| +// the current state of the cache is too old to determine whether the record | |
| +// will apply. In such cases the cache should be refreshed to recordAliveAt and | |
| +// then this method should be called again. | |
| +// In certain cases we return a doesNotApplyReason explaining why the protected | |
| +// ts record does not currently apply. We do not want to return an error so that | |
| +// we can aggregate the reasons across multiple | |
| +// AdminVerifyProtectedTimestampRequest, as explained in | |
| +// adminVerifyProtectedTimestamp. | |
| +func (r *Replica) protectedTimestampRecordCurrentlyApplies( | |
| + ctx context.Context, args *roachpb.AdminVerifyProtectedTimestampRequest, | |
| +) (willApply, cacheTooOld bool, doesNotApplyReason string, _ error) { | |
| + // We first need to check that we're the current leaseholder. | |
| + // TODO(ajwerner): what other conditions with regards to time do we need to | |
| + // check? I don't think there are any. If the recordAliveAt is after our | |
| + // liveness expiration that's okay because we're either going to find the | |
| + // record or we're not and if we don't then we'll push the cache and re-assert | |
| + // that we're still the leaseholder. If somebody else becomes the leaseholder | |
| + // then they will have to go through the same process. | |
| + ls, pErr := r.redirectOnOrAcquireLease(ctx) | |
| + if pErr != nil { | |
| + return false, false, "", pErr.GoError() | |
| } | |
| - var protectionTimestamps []hlc.Timestamp | |
| - var err error | |
| - protectionTimestamps, ts.readAt, err = r.store.protectedtsReader.GetProtectionTimestamps(ctx, sp) | |
| - if err != nil { | |
| - return ts, err | |
| - } | |
| - earliestTS := hlc.Timestamp{} | |
| - for _, protectionTimestamp := range protectionTimestamps { | |
| - // Check if the timestamp the record was trying to protect is strictly | |
| - // below the GCThreshold, in which case, we know the record does not apply. | |
| - if isValid := gcThreshold.LessEq(protectionTimestamp); !isValid { | |
| - continue | |
| - } | |
| - if earliestTS.IsEmpty() || protectionTimestamp.Less(earliestTS) { | |
| - earliestTS = protectionTimestamp | |
| + | |
| + // NB: It should be the case that the recordAliveAt timestamp | |
| + // is before the current time and that the above lease check means that | |
| + // the replica is the leaseholder at the current time. If recordAliveAt | |
| + // happened to be newer than the current time we'd need to make sure that | |
| + // the current Replica will be live at that time. Given that recordAliveAt | |
| + // has to be before the batch timestamp for this request and we should | |
| + // have forwarded the local clock to the batch timestamp this can't | |
| + // happen. | |
| + // TODO(ajwerner): do we need to assert that indeed the recordAliveAt precedes | |
| + // the batch timestamp? Probably not a bad sanity check. | |
| + | |
| + // We may be reading the protected timestamp cache while we're holding | |
| + // the Replica.mu for reading. If we do so and find newer state in the cache | |
| + // then we want to, update the replica's cache of its state. The guarantee | |
| + // we provide is that if a record is successfully verified then the Replica's | |
| + // cachedProtectedTS will have a readAt value high enough to include that | |
| + // record. | |
| + var read cachedProtectedTimestampState | |
| + defer r.maybeUpdateCachedProtectedTS(&read) | |
| + r.mu.RLock() | |
| + defer r.mu.RUnlock() | |
| + defer read.clearIfNotNewer(r.mu.cachedProtectedTS) | |
| + | |
| + // If the key that routed this request to this range is now out of this | |
| + // range's bounds, return an error for the client to try again on the | |
| + // correct range. | |
| + desc := r.descRLocked() | |
| + if !kvserverbase.ContainsKeyRange(desc, args.Key, args.EndKey) { | |
| + return false, false, "", roachpb.NewRangeKeyMismatchErrorWithCTPolicy(ctx, args.Key, args.EndKey, desc, | |
| + r.mu.state.Lease, r.closedTimestampPolicyRLocked()) | |
| + } | |
| + if args.Protected.LessEq(*r.mu.state.GCThreshold) { | |
| + gcReason := fmt.Sprintf("protected ts: %s is less than equal to the GCThreshold: %s for the"+ | |
| + " range %s - %s", args.Protected.String(), r.mu.state.GCThreshold.String(), | |
| + desc.StartKey.String(), desc.EndKey.String()) | |
| + return false, false, gcReason, nil | |
| + } | |
| + if args.RecordAliveAt.Less(ls.Lease.Start.ToTimestamp()) { | |
| + return true, false, "", nil | |
| + } | |
| + | |
| + // Now we're in the case where maybe it is possible that we're going to later | |
| + // attempt to set the GC threshold above our protected point so to prevent | |
| + // that we add some state to the replica. | |
| + r.protectedTimestampMu.Lock() | |
| + defer r.protectedTimestampMu.Unlock() | |
| + if args.Protected.Less(r.protectedTimestampMu.pendingGCThreshold) { | |
| + gcReason := fmt.Sprintf( | |
| + "protected ts: %s is less than the pending GCThreshold: %s for the range %s - %s", | |
| + args.Protected.String(), r.protectedTimestampMu.pendingGCThreshold.String(), | |
| + desc.StartKey.String(), desc.EndKey.String()) | |
| + return false, false, gcReason, nil | |
| + } | |
| + | |
| + var seen bool | |
| + read = r.readProtectedTimestampsRLocked(ctx, func(r *ptpb.Record) { | |
| + // Comparing record ID and the timestamp ensures that we find the record | |
| + // that we are verifying. | |
| + // A PTS record can be updated with a new Timestamp to protect, and so we | |
| + // need to ensure that we are not seeing the old version of the record in | |
| + // case the cache has not been updated. | |
| + if r.ID.GetUUID() == args.RecordID && args.Protected.LessEq(r.Timestamp) { | |
| + seen = true | |
| } | |
| + }) | |
| + | |
| + // If we observed the record in question then we know that all future attempts | |
| + // to run GC will observe the Record if it still exists. The one hazard we | |
| + // need to avoid is a race whereby an attempt to run GC first checks the | |
| + // protected timestamp state and then attempts to increase the GC threshold. | |
| + // We set the minStateReadTimestamp here to avoid such races. The MVCC GC | |
| + // queue will call markPendingGC just prior to sending a request to update the | |
| + // GC threshold which will verify the safety of the new value relative to | |
| + // minStateReadTimestamp. | |
| + if seen { | |
| + r.protectedTimestampMu.minStateReadTimestamp = read.readAt | |
| + return true, false, "", nil | |
| + } | |
| + | |
| + isCacheTooOld := read.readAt.Less(args.RecordAliveAt) | |
| + // Protected timestamp state has progressed past the point at which we | |
| + // should see this record. This implies that the record has been removed. | |
| + if !isCacheTooOld { | |
| + recordRemovedReason := "protected ts record has been removed" | |
| + return false, false, recordRemovedReason, nil | |
| } | |
| - ts.earliestProtectionTimestamp = earliestTS | |
| - return ts, nil | |
| + // Retry, since the cache is too old. | |
| + return false, true, "", nil | |
| } | |
| // checkProtectedTimestampsForGC determines whether the Replica can run GC. If | |
| @@ -108,7 +261,7 @@ func (r *Replica) readProtectedTimestampsRLocked( | |
| // old gc threshold, and the new gc threshold. | |
| func (r *Replica) checkProtectedTimestampsForGC( | |
| ctx context.Context, gcTTL time.Duration, | |
| -) (canGC bool, cacheTimestamp, gcTimestamp, oldThreshold, newThreshold hlc.Timestamp, _ error) { | |
| +) (canGC bool, cacheTimestamp, gcTimestamp, oldThreshold, newThreshold hlc.Timestamp) { | |
| // We may be reading the protected timestamp cache while we're holding | |
| // the Replica.mu for reading. If we do so and find newer state in the cache | |
| @@ -127,16 +280,12 @@ func (r *Replica) checkProtectedTimestampsForGC( | |
| // read.earliestRecord is the record with the earliest timestamp which is | |
| // greater than the existing gcThreshold. | |
| - var err error | |
| - read, err = r.readProtectedTimestampsRLocked(ctx) | |
| - if err != nil { | |
| - return false, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{}, err | |
| - } | |
| + read = r.readProtectedTimestampsRLocked(ctx, nil) | |
| gcTimestamp = read.readAt | |
| - if !read.earliestProtectionTimestamp.IsEmpty() { | |
| + if read.earliestRecord != nil { | |
| // NB: we want to allow GC up to the timestamp preceding the earliest valid | |
| - // protection timestamp. | |
| - impliedGCTimestamp := gc.TimestampForThreshold(read.earliestProtectionTimestamp.Prev(), gcTTL) | |
| + // record. | |
| + impliedGCTimestamp := gc.TimestampForThreshold(read.earliestRecord.Timestamp.Prev(), gcTTL) | |
| if impliedGCTimestamp.Less(gcTimestamp) { | |
| gcTimestamp = impliedGCTimestamp | |
| } | |
| @@ -145,12 +294,12 @@ func (r *Replica) checkProtectedTimestampsForGC( | |
| if gcTimestamp.Less(lease.Start.ToTimestamp()) { | |
| log.VEventf(ctx, 1, "not gc'ing replica %v due to new lease %v started after %v", | |
| r, lease, gcTimestamp) | |
| - return false, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{}, nil | |
| + return false, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{}, hlc.Timestamp{} | |
| } | |
| newThreshold = gc.CalculateThreshold(gcTimestamp, gcTTL) | |
| - return true, read.readAt, gcTimestamp, oldThreshold, newThreshold, nil | |
| + return true, read.readAt, gcTimestamp, oldThreshold, newThreshold | |
| } | |
| // markPendingGC is called just prior to sending the GC request to increase the | |
| diff --git a/pkg/kv/kvserver/replica_protected_timestamp_test.go b/pkg/kv/kvserver/replica_protected_timestamp_test.go | |
| index f02ca7af3e..31c54420f8 100644 | |
| --- a/pkg/kv/kvserver/replica_protected_timestamp_test.go | |
| +++ b/pkg/kv/kvserver/replica_protected_timestamp_test.go | |
| @@ -12,23 +12,397 @@ package kvserver | |
| import ( | |
| "context" | |
| - "math/rand" | |
| + "fmt" | |
| "testing" | |
| "time" | |
| "github.com/cockroachdb/cockroach/pkg/keys" | |
| + "github.com/cockroachdb/cockroach/pkg/kv/kvserver/pro |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment