Capistrano Resque

Basic tasks for putting some Resque in your Cap. This should be fully compatible with both Capistrano 2.x and 3.x, but if you run into any issues please report them.

At this time, we are only targeting Resque 1.x; the 2.0/master branch is still a work-in-progress without a published gem.

In your Gemfile:

gem "capistrano-resque", "~> 0.2.2", require: false

In lib/tasks:

You'll need to make sure your app is set to include Resque's rake tasks. Per the Resque 1.x README, you'll need to add require 'resque/tasks' somewhere under the lib/tasks directory (e.g. in a lib/tasks/resque.rake file).

In your Capfile:

Put this line after any of capistrano's own require/load statements (specifically load 'deploy' for Cap v2):

require "capistrano-resque"

Note: You must tell Bundler not to automatically require the file (by using require: false), otherwise the gem will try to load the Capistrano tasks outside of the context of running the cap command (e.g. running rails console).

In your deploy.rb:

# Specify the server that Resque will be deployed on. If you are using Cap v3
# and have multiple stages with different Resque requirements for each, then
# these __must__ be set inside of the applicable config/deploy/... stage files
# instead of config/deploy.rb:
role :resque_worker, "app_domain"
role :resque_scheduler, "app_domain"

set :workers, { "my_queue_name" => 2 }

# We default to storing PID files in a tmp/pids folder in your shared path, but
# you can customize it here (make sure to use a full path). The path will be
# created before starting workers if it doesn't already exist.
# set :resque_pid_path, -> { File.join(shared_path, 'tmp', 'pids') }

# Uncomment this line if your workers need access to the Rails environment:
# set :resque_environment_task, true

You can also specify multiple queues and the number of workers for each queue:

set :workers, { "archive" => 1, "mailing" => 3, "search_index, cache_warming" => 1 }

The above will start five workers in total:

• one listening on the archive queue
• one listening on the search_index, cache_warming queue
• three listening on the mailing queue

If you need to pass arbitrary data (like other non-standard environment variables) to the "start" command, you can specify:

set :resque_extra_env, "SEARCH_SERVER=172.18.0.52"

This can be useful for customizing Resque tasks in complex server environments.

Multiple Servers/Roles

You can also start up workers on multiple servers/roles:

role :worker_server_A,  <server-ip-A>
role :worker_servers_B_and_C,  [<server-ip-B>, <server-ip-C>]

set :workers, {
  worker_server_A: {
    "archive" => 1,
    "mailing" => 1
  },
  worker_servers_B_and_C: {
    "search_index" => 1,
  }
}

The above will start four workers in total:

• one archive on Server A
• one mailing on Server A
• one search_index on Server B
• one search_index on Server C

Rails Environment

With Rails, Resque requires loading the Rails environment task to have access to your models, etc. (e.g. QUEUE=* rake environment resque:work). However, Resque is often used without Rails (and even if you are using Rails, you may not need/want to load the Rails environment). As such, the environment task is not automatically included.

If you would like to load the environment task automatically, add this to your deploy.rb:

set :resque_environment_task, true

If you would like your workers to use a different Rails environment than your actual Rails app:

set :resque_rails_env, "my_resque_env"

The tasks

Running cap -vT | grep resque should give you...

➔ cap -vT | grep resque
cap resque:status    # Check workers status
cap resque:start     # Start Resque workers
cap resque:stop      # Quit running Resque workers
cap resque:restart   # Restart running Resque workers
cap resque:scheduler:restart #
cap resque:scheduler:start   # Starts Resque Scheduler with default configs
cap resque:scheduler:stop    # Stops Resque Scheduler

Restart on deployment

To restart you workers automatically when cap deploy:restart is executed add the following line to your deploy.rb:

after "deploy:restart", "resque:restart"

Logging

Backgrounding and logging are current sticking points. I'm using the HEAD of resque's 1-x-stable branch for the 0.0.8 release because it has some new logging functions not yet slated for a resque release.

In your Gemfile, you will need to specify:

gem 'resque', :git => 'git://github.com/resque/resque.git', :branch => '1-x-stable'

Also, you will need to include:

Resque.logger = Logger.new("new_resque_log_file")

...somewhere sensible, such as in your resque.rake, to achieve logging.

The chatter on: https://github.com/defunkt/resque/pull/450 gives more information. If using HEAD of this resque branch doesn't work for you, then pin to v0.0.7 of this project.

Redirecting output

Due to issues in the way Resque 1.x handles background processes, we automatically redirect stderr and stdout to /dev/null.

If you'd like to capture this output instead, just specify a log file:

set :resque_log_file, "log/resque.log"

You can also disable the VERBOSE option to reduce the amount of log output:

set :resque_verbose, false

Limitations

Starting workers is done concurrently via Capistrano and you are limited by ssh connections limit on your server (default limit is 10)

To to use more workers, please change your sshd configuration (/etc/ssh/sshd_config)

MaxStartups 100

Contributing

1. Fork it
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Added some feature')
4. If possible, make sure your changes apply to both the Capistrano v2 and v3 code (capistrano_integration.rb is v2, capistrano-resque.rake is v3)
5. Push to the branch (git push origin my-new-feature)
6. Create new Pull Request

License

Please see the included LICENSE file.