Using inotify to monitor system file changed in a directory and take action.

optawatch.py

#!/usr/bin/env python

import re
import os
import logging
import argparse
import pyinotify
import subprocess


def get_args():
    parser = argparse.ArgumentParser()

    parser.add_argument("--directory", dest="dirname", required=True,
        help="Give the directory to monitor, ex: /tmp")

    return parser.parse_args()


class EventHandler(pyinotify.ProcessEvent):
    def process_IN_CLOSE_WRITE(self, event):
        try:
            p = re.compile('.*\.tmp$')
            if not p.search(event.pathname):
                logging.info("Opta write event: %s" % event.pathname)
                {%- for hostname in pillar['opta_remote_hosts'] %}
                cmd = ['/usr/bin/rsync', '-tdz', '--delete-after', '/home/opta/stats/', '{{ hostname }}:/home/opta/stats/']
                p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
                stdout_value, stderr_value = p.communicate()
                if stderr_value:
                    logging.error(stderr_value)
                elif stdout_value:
                    logging.info(stdout_value)
                {%- endfor %}
        except sh.ErrorReturnCode_12, err:
            logging.debug(err)
        except Exception, err:
            logging.debug(err)

    def process_IN_DELETE(self, event):
        try:
            p = re.compile('.*\.tmp$')
            if not p.search(event.pathname):
                logging.info("Opta delete event: %s" % event.pathname)
                {%- for hostname in pillar['opta_remote_hosts'] %}
                cmd = ['/usr/bin/rsync', '-tdz', '--delete-after', '/home/opta/stats/', '{{ hostname }}:/home/opta/stats/']
                p = subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
                stdout_value, stderr_value = p.communicate()
                if stderr_value:
                    logging.error(stderr_value)
                elif stdout_value:
                    logging.info(stdout_value)
                {%- endfor %}
        except sh.ErrorReturnCode_12, err:
            logging.debug(err)
        except Exception, err:
            logging.debug(err)


def run():
    logging.basicConfig(filename='/var/log/optawatch.log',level=logging.DEBUG)

    args = get_args()

    pid_file = "/tmp/optawatch.pid"

    try:
        open(pid_file, 'r')
        os.remove(pid_file)
    except IOError:
        if args.dirname:
            wm = pyinotify.WatchManager()
            mask = pyinotify.IN_CLOSE_WRITE | pyinotify.IN_DELETE # watched events
            handler = EventHandler()
            notifier = pyinotify.Notifier(wm, handler)
            wdd = wm.add_watch(args.dirname, mask, rec=True)

            try:
                notifier.loop(daemonize=True,
                    pid_file=pid_file, stdout='/tmp/stdout.txt')
            except pyinotify.NotifierError, err:
                logging.debug(err)
        else:
            logging.warning("You must specify an existing directory to monitor.")
    except Exception, err:
        logging.debug("Unexplained error - %s" % err)

if __name__ == "__main__":
    run()

This process is no longer being used as rsync was taking up too many resources. This started out as a 1:1 copy of an existing script in perl. The next incarnation uses paraminko, the ssh module, to directly handle the transfers.