How can I automate resque workers to update their code when I push my app to heroku?

Question

The problem is:

When I make changes on the code performed by the workers, the running workers on heroku still executes with the previous after I pushed a new code.

Then I have to manually stop and start again resque. But its wrong, I guess, because the worker could be performing some job and must have to wait it until fininhes to be unregistered.

This is my resque.rake

require 'resque'
require "resque/tasks"

task "resque:setup" => :environment do
  ENV['QUEUE'] = '*' if ENV['QUEUE'].blank?

  Resque::Worker.all.each {|w| w.unregister_worker}
  Resque.after_fork do |job| 
    ActiveRecord::Base.establish_connection
  end
end

task "resque:clean_workers" => :environment do
  Resque::Worker.all.each {|w| w.unregister_worker}
end

desc "Alias for resque:work (To run workers on Heroku)"
task "jobs:work" => "resque:work"

Dir["#{Rails.root}/app/jobs/*.rb"].each { |file| require file }

I would like to know if when I push my app to heroku, the worker could be updated with the new code. How could it be done?

Thanks!!

Answer 1

Heroku restarts all your processes on deploy as a platform feature -- except for one-off processes spawned via heroku run .

Add the worker process type to your Procfile :

worker: bundle exec rake resque:work QUEUE=*

Push, and then tell Heroku to run your worker:

$ heroku ps:scale worker=1

Heroku will now ensure that you have a single worker dyno running at all times. It'll restart it automatically when your app restarts (eg after a push, a heroku restart , a rollback ), or if your worker dies somehow, or if the host it's on fails. Heroku will even gracefully shut down and restart your worker process on a new host every day or so just for good measure, purging temp files, putting a cap on any memory leaks, and rebalancing load around the dyno grid.

This is the behavior you want if you're running background jobs 24x7 in production.