cronify 0.1

Event based cron-like utility using the linux kernels' inotify API

Cronify is an event based task scheduler that uses the Linux kernels’ inotify API.

It allows for triggering of actions when files appear in a directory, with configurable action parameters that can be file data such as file name and file date stamp and for scheduling actions to be run within a certain time period among other things.

The service’s configuration is YAML based. You may of course run cronify as a python module as well.

Once installed, it is run as a service, or daemon, much like cron itself.

Installation

Requires a running Linux kernel >=2.6.13

$ sudo pip install cronify
$ sudo chmod +x /etc/init.d/cronifyd
# Install a /etc/cronify.yaml file as in the examples below
# Once a configuration file is in place, start the cronify daemon with the provided init script
$ sudo /etc/init.d/cronifyd start

Configuration Example

$ cat /etc/cronify.yaml

/tmp/testdir :
    name : Access log watcher
    recurse : false
    filemasks :
      somefile.* :
        actions :
          - processFile :
              args:
                - $filename
                - YYYYMMDD
              cmd: echo``

$ sudo /etc/init.d/cronifyd start
$ touch /tmp/testdir/somefile.txt
$ tail /var/log/cronify/cronify.log

cronify.cronify - Thread-1 - 2013-03-26 17:40:40,485 - INFO - Got result from action {'cmd': 'echo', 'args': ['echo', '/tmp/testdir/somefile.txt', '20130326']} - /tmp/testdir/somefile.txt 20130326

More complex configuration with multiple watchers and delayed actions

See example.yaml in repository for complete list of accepted configuration

$ cat /etc/cronify.yaml

/tmp/testdir :
    name : Access log watcher
    recurse : false
    filemasks :
      access_log_YYYYMMDD.* :
        actions :
          - processFile :
              args:
                - $filename
                - YYYYMMDD
              cmd: process

/tmp/testdir2 :
    name : Other log watcher
    recurse : true
    filemasks :
      other_log_YYYYMMDD.* :
         actions :
          # Actions to perform on the file in sequence.
          - processFile :
            # Do not start action before this time. Action is queued until start_time if triggered prior to it.
            # This configuration setting is optional
            start_time: 0800
            # Do not start action after this time. Action is queued until next start_time if triggered after end time
            # This configuration setting is optional
            end_time: 1000
            args:
              - $filename
              - YYYYMMDD
            cmd: process

Known limitations

• Currently queued actions will be _lost_ upon a service restart.

• Reloading of the cronify configuration file is not yet possible without a restart.

• When using recurse, inotify is limited to watching N number of subdirectories in the tree, where N is value of /proc/sys/fs/inotify/max_user_watches. See http://linux.die.net/man/7/inotify

  User can increase this limit by modifying /proc/sys/fs/inotify/max_user_watches

• When watching an NFS directory on NFS server side, only events made by the NFS server will be seen by the inotify API and following, cronify itself.

  When watching an NFS directory on NFS client side, no events are seen by inotify at all.

  In other words if you were planning on watching for a file that is created by an NFS client, this is currently not possible.