resque-multi-step

Resque multi-step provides an abstraction for managing multiple step async tasks.

This software is not considered stable at this time. Use at your own risk.

Consider a situation where you need to perform several actions and would like those actions to run in parallel and you would like to keep track of the progress. Mutli-step can be use to implement that as follows:

task = Resque::Plugins::MultiStepTask.create("pirate-take-over") do |task|
  blog.posts.each do |post|                                     
    task.add_job ConvertPostToPirateTalk, post.id
  end
end

A resque job will be queued for each post. The task object will keep track of how many of the tasks have been completed successfully (#completed_count). That combined with the overall job count (#total_job_count) make it easy to compute the percentage completion of a mutli-step task.

The failed job count (#failed_count) makes it easy to determine if problem has occurred during the execution.

Once you have kicked off a job you can look it up again later using it's task id. First you persist the task id when you create the task.

task = Resque::Plugins::MultiStepTask.create('pirate-take-over") do |task|
  ...
end
blog.async_task_id = task.task_id
blog.save!

Then you can look it up using the .find method on MultiStepTask.

# Progress reporting action; executed in a different process.
begin
  task = Resque::Plugins::MultiStepTask.find(blog.async_task_id)
  render :text => "percent complete #{(task.completed_count.quo(task.total_job_count) * 100).round}%
  
rescue Resque::Plugins::MultiStepTask::NoSuchMultiStepTask
  # task completed...
 
  redirect_to blog_url(blog)
end

As of version 2.0.0, you can also get a handle on the original multi-step task. This is useful if you need to add another job on the fly, that needs to happen before the finalization.

class ConvertPostToPirateTalk
  def self.perform(post_id)
    p = Post.find(post_id)
    p.convert_to_pirate_talk
    p.save
    
    if p.has_comments?
      p.comments.each do |c|
        # #multi_step_task is a class method defined on the class
        # before invoking #perform as a convenience. you can also
        # use #multi_step_task_id to just get the resque key.
        multi_step_task.add_job ConvertCommentToPirateTalk, c.id
      end
    end
  end
end

Often when doing mutli-step tasks there are a bunch of tasks that can all happen in parallel and then a few that can only be executed after all the rest have completed. Mutli-step task finalization supports just that use case.

Using our example, say we want to commit the solr index and then unlock the blog we are converting to pirate talk once the conversion is complete.

task = Resque::Plugins::MultiStepTask.create("pirate-take-over") do |task|
  blog.posts.each do |post|                                     
    task.add_job ConvertPostToPirateTalk, post.id
  end
  
  task.add_finalization_job CommitSolr
  task.add_finalization_job UnlockBlog, blog.id
end    

This would convert all the posts to pirate talk in parallel, using as many workers as are available. Once all the normal jobs are completed the finalization jobs are run serially in a single worker. Finalization are executed in the order in which they are registered. In our example, solr will be committed and then, after the commit is complete, the blog will be unlocked.

MultiStepTask creates a queue in resque for each task. To process multi-step jobs you will need at least one Resque worker with QUEUES=*. This combined with [resque-fairly][] provides fair scheduling of the constituent jobs.

Having a queue per multi-step task means that is easy to determine to what task a particular job belongs. It also provides a nice way to see what is going on in the system at any given time. Just got to resque-web and look the queue list. Use meaningful slugs for your tasks and you get a quick birds-eye view of what is going on.

As of version 2.0.0 MultiStepTask requires ruby 1.9 or higher.

• Fork the project.
• Make your feature addition or bug fix.
• Add tests for it. This is important so I don't break it in a future version unintentionally.
• Update history to reflect the change, increment the version properly as described at http://semver.org/.
• Commit
• Send me a pull request. Bonus points for topic branches.

To join the list simply send an email to mailto:resquemultistep@librelist.com. This will subscribe you and send you information about your subscription, including unsubscribe information.

The archive can be found at http://librelist.com/browser/resquemultistep/.

Copyright (c) OpenLogic, Inc. See LICENSE for details.