The goal to deal with one problem: The startup time of a big Rails project when running a functional suite across multiple servers.

The project this lil' app was started on is running a Cucumber suite that spans a total of 10 concurrent runners, all of whom need data at different times. It was originally developed with rake tasks that was reused from local development and run on the servers remotely through SSH.

But ~30s startup time per task was starting to have too big of an effect, so this gem was born with the hope of removing all the starting cost while keeping it really simple to reuse the already existing rake tasks.

Add to your Gemfile and ensure you mount the app in your routes.

To your Gemfile:

gem 'remote_rails_rake_runner'

To your routes.rb:

Rails.application.routes.draw do

  mount RemoteRailsRakeRunner::Engine => '/rake'
end

You can specify the path to your Rakefile by setting the option Rails.application.config.remote_rake_runner_rakefile_path which defaults to Rails.root.join('Rakefile').to_s.

The app exposes two endpoints, a listing of all the rake tasks and the actual runner. To run a task with arguments just put the arguments in into the argument args separated by commas. The task will not be considered successful if it raises an exception, otherwise it's assumed to have been successful.

If you mount the app under /rake as suggested above:

$ curl http://localhost:3000/rake
[{"name":"about","args":[],"description":null}, …] # a listing of all tasks (like rake -T)

$ curl http://localhost:3000/rake/simple:hello -d args=Björn
{"success":true,"output":"Hello Björn!\n"}

The running and capturing of output that was taken from this blog post by Andrea Ercolino.

This project rocks and uses MIT-LICENSE.