Unicorn Stack

Intercity chef-repo supports two ways of running your application code on your server:

• Nginx + Phusion Passenger (preferred in the README and samples)
• Nginx + Unicorn

We prefer to use Phusion Passenger as the default stack because it is easier to integrate, update and deploy applications. However, we also fully support using Unicorn to run your Rails code.

The steps on this page show you how to use Unicorn instead of Phusion Passenger.

Changing the node json file

To have our cookbooks configure Unicorn as Rails application runner instead of Passenger, change role[rails_passenger] to role[rails] in the run_list key of your host .json configuration.

So to get Unicorn with MySQL you set:

{
  "run_list": ["role[mysql]", "role[rails]"],
  ... other configuration

And for Unicorn with PostgreSQL you set:

{
  "run_list": ["role[postgresql]", "role[rails]"],
  ... other configuration

Configuring Capistrano

Add the capistrano3-unicorn gem to your Gemfile:

gem "capistrano3-unicorn", "~> 0.2.1", group: :development

and run bundle.

In your Capfile add the line:

require 'capistrano3/unicorn'

In config/deploy.rb add these lines:

set :unicorn_config_path, "#{current_path}/config/unicorn.rb"
set :bundle_bins, fetch(:bundle_bins, []).push("unicorn")

And add unicorn.rb to the :linked_files setting so it looks like:

set :linked_files, %w{config/database.yml .rbenv-vars config/unicorn.rb}

Also in config/deploy.rb replace the full namespace :deploy do with:

namespace :deploy do
  task :restart do
    invoke 'unicorn:restart'
  end

  after 'deploy:publishing', 'restart'
end

That's it! You can now deploy your app and start/reload Unicorn with:

cap production deploy