a sample script that sleeps

flocking

# round 1! => nonblocking example
[process a] $ flock -en sleeper.lock ./sleeper.sh
time to sleep ...
[process b] $ flock -en sleeper.lock ./sleeper.sh
# ^ will exit b/c of n switch: "Fail (with an exit code of 1) 
#   rather than wait if the lock cannot be immediately acquired.
... waking up # <= process a returning


# round 2! => blocking example
[process a] $ flock -en sleeper.lock ./sleeper.sh
time to sleep ...
[process b] $ flock -e sleeper.lock ./sleeper.sh
# ^ will wait for the lock to become available b/c not using -n this time
... waking up # <= process a returning
time to sleep ... # <= process b executing sleeper.sh
... waking up # <= process b returning


# round 3! => different lock example
[process a] $ flock -en sleeper.lock ./sleeper.sh
time to sleep ...
[process b] $ flock -en different-sleeper.lock ./sleeper.sh
time to sleep ...
... waking up # <= process a returning
... waking up # <= process b returning
# ^ process b runs without waiting because flock is not using the same
#   lock file to ensure a lock. Take away: lockfile names matter!


# round 4! => locks are advisory only
[process a] $ flock -en sleeper.lock ./sleeper.sh
time to sleep ...
[process b] $ ./sleeper.sh
time to sleep ...
... waking up # <= process a returning
... waking up # <= process b returning
# ^ process b runs without waiting because it didn't use flock
#   locking is only advisory and will not do anything if
#   flock is not always used.

sleeper.sh

#!/bin/bash
# All this script does it announce it's going to sleep,
# sleeps, and then announces when it wakes back up
echo "time to sleep ..."
sleep 10
echo "... waking up"