rescue Resque

Chiel Wester wo 14 jul 10

A couple of months ago, i posted an article on this blog about implementing resque for performing background jobs.

There have been several updates to the resque core since then, one of them is the possibility to write plugins for resque without having to change the main functionality of resque.

One of these plugins that was created is resque-retry. When you include this plugin in your Job class, any failed job will be retried several times when it raises an error. This becomes very useful when your job uses some external service that can give the pain of timeouts.

The plugin also has the possibility to only retry on specific exceptions (e.g. Timeout, NetworkError). At the moment the job is simply deleted when the retry limit is exceeded. Of course you could overwrite this functionality to notify you in some way that a job failed too many times:

    def on_failure_retry(exception, *args)
      if retry_criteria_valid?(exception, *args)
        try_again(*args)
      else
        # Notify me about this error!
        Resque.redis.del(redis_retry_key(*args))
      end
    end

Check for examples and documentation the github project page

This plugin depends on the resque-scheduler plugin that gives you the possibility to schedule jobs in a cron-like way or to schedule a job at some time in the future.

Gepost in hor |  0 reacties