Deploying Rails 4.1 apps with Resque to Heroku

June 2, 2014 | technical, rails, heroku, resque

The Heroku guides for deploying Rails apps encourage you to do so using a Procfile and Unicorn. Not being super deployment savvy, I tend to follow the instructions provided to me.

… that is, until they completely, totally, and utterly fail me.

After I added Resque to a Rails app I was working on, I was having a hell of a time deploying it to Heroku: first Redis wouldn’t start, and then when Redis finally started, my workers wouldn’t start. Nothing in the guides for Resque, or two Redis add ons (RedisToGo and Redis Cloud), was helpful or even remotely close to correct.

Some intense Googling turned up some issues related to older versions of Resque (1.22.1, which Heroku uses in their guides) and some compatibility issues with Unicorn.

I don’t think the compatibility issues have been resolved, but I was able to come up with a fix that works locally and on production without any hiccups.

I am using Redis Cloud in my configuration below, but this should work with other Redis add ons.


This guide works specifically with the following versions of Rails, Unicorn, and Resque:

gem 'rails', '4.1.1'
gem 'resque', '~> 1.24.1'
gem 'unicorn', '~> 4.6.2'

I can’t make any promises about other versions.


This is pretty standard for Heroku, Resque, and Rails.

web: bundle exec unicorn -p $PORT -c ./config/unicorn.rb
resque: env TERM_CHILD=1 QUEUE=* bundle exec rake resque:work


Create a new file: config/initializers/redis.rb and put this inside of it:

  $redis = Resque.redis = => ENV["REDISCLOUD_URL"])


We need access to the Rake tasks. Create a new file: lib/tasks/resque.rake.

require 'resque/tasks'

task "resque:preload" => :environment


The last thing we need to do is configure Unicorn, which turned out to be the most difficult part.

In your config/unicorn.rb file, put:

worker_processes Integer(ENV["WEB_CONCURRENCY"] || 3)
timeout 15
preload_app true

before_fork do |server, worker|
  Signal.trap 'TERM' do
    puts 'Unicorn master intercepting TERM and sending myself QUIT instead'
    Process.kill 'QUIT',

  defined?(ActiveRecord::Base) and

  if defined?(Resque)
    Resque.redis.quit'Disconnected from Redis')

after_fork do |server, worker|
  Signal.trap 'TERM' do
    puts 'Unicorn worker intercepting TERM and doing nothing. Wait for master to send QUIT'

  if defined?(Resque)'Connected to Redis')

And you should be all set to deploy to Heroku with ease!