Generate statistics for Cloudwatch based on the number of SpamAssassin Childs activity

sachilds.py

# -*- config:utf-8 -*-
import boto3, asyncio, sys, os, datetime


def get_sa_max_children(default=12):
    try:
        options = None
        with open('/etc/default/spamassassin', 'r') as f:
            for line in f:
                if line.find('OPTIONS=') == 0:
                    options = line.strip()
                    break

        args = iter(options[9:].split(' '))
        for arg in args:
            if arg[0:4] == '--max-children=':
                return int(arg[15:])
            elif arg == '-m':
                return int(next(args, None))
    except Exception:
        pass

    return default


def get_activity(idles):
    block = 1024
    log_file = '/var/log/mail.info'
    file_size = os.stat(log_file)[6]
    start_position = None
    with open(log_file, 'r') as f:
        f.seek(file_size, 0)  # Go to end
        data = ''
        for i in range(1, 100):
            # We go up until we find the pattern
            f.seek((file_size - (block * i)), 0) # we go back {block} from the current position
            part = f.read(block)
            data = part + data
            start_position = data.rfind('prefork: child states:')
            if start_position > 60:
                # This is to ensure the position is after the last line
                break

    if start_position is None:
        return None, None

    state = None
    for line in data.split('\n')[::-1]:
        if line.find('prefork: child states:') > -1:
            before, after = line.split('prefork: child states:', 1)
            parsed_date = None
            try:
                # Trying date format 'Wed Nov  1 17:10:39 2023'
                dt = before[0:before.find(' [')].strip()
                parsed_date = datetime.datetime.strptime(dt, "%a %b %d %H:%M:%S %Y")
            except ValueError:
                # Might be 'Nov  1 17:11:44'
                dt = before[0:before.find(' [')].strip()
                dt = dt[0:dt.rfind(' ')] # remove spamd[xxx]
                dt = dt[0:dt.rfind(' ')] # remove ip-XXX
                try:
                    parsed_date = datetime.datetime.strptime(dt, "%b %d %H:%M:%S")
                    parsed_date = parsed_date.replace(year=datetime.datetime.now().year)
                except ValueError:
                    pass

            if parsed_date and parsed_date < datetime.datetime.now() - datetime.timedelta(minutes=15):
                return None, None

            state = after.strip()
            break

    busy = 0
    for child in list(state):
        if child == 'B':
            busy += 1
            idles -= 1

    activity = 100 if busy > idles else (busy / idles) * 100

    print('')
    print('Idles: {}'.format(idles))
    print('Busy: {}'.format(busy))
    print('Activity: {}%'.format(activity))

    return idles, busy, activity


async def main(debug=False, is_delayed=False):
    default_idles = get_sa_max_children()

    if debug:
        return get_activity(default_idles)

    for i in range(0, 5):
        idles, busy, activity = get_activity(default_idles)

        try:
            assert busy is not None
            assert activity is not None

            boto3.client('cloudwatch', region_name='eu-west-3').put_metric_data(
                Namespace='SpamAssassin Delayed statistics' if is_delayed else 'SpamAssassin statistics',
                MetricData=[
                    {
                        'MetricName': 'Idle',
                        'Unit': 'Count',
                        'Value': idles,
                        'StorageResolution': 1,
                        'Timestamp': datetime.datetime.utcnow()
                    },
                    {
                        'MetricName': 'Busy',
                        'Unit': 'Count',
                        'Value': busy,
                        'StorageResolution': 1,
                        'Timestamp': datetime.datetime.utcnow()
                    },
                    {
                        'MetricName': 'Activity',
                        'Unit': 'Percent',
                        'Value': activity,
                        'StorageResolution': 1,
                        'Timestamp': datetime.datetime.utcnow()
                    },
                ]
            )
        except Exception:
            # An error occured
            pass

        await asyncio.sleep(10) # We wait 10 seconds


if __name__ == '__main__':
    is_debug = '--debug' in sys.argv
    is_delayed = '--delayed' in sys.argv
    asyncio.get_event_loop().run_until_complete(main(is_debug, is_delayed))