unicorn-ng Cookbook

Manage your unicorn application server with this cookbook.

This cookbook provides a decent initscript for Debian/Ubuntu, as well as takes care of the unicorn configuration.

Features

• Provides a decent initscript for Debian/Ubuntu
• Automatically takes care of ActiveRecord connections (if ActiveRecord is used)
• Hightly configurable for your needs
• Use either a system-wide installed unicorn, or one that gets deployed with your application

Requirements

An installed unicorn. Alterantively you can use the install recipe to install it using rubygems.

Furthermore you need to add the following line to your metadata.rb

depends 'unicorn-ng'

Attributes

Packages

To install unicorn, rubygems is required. The cookbook tries to figure out automatically which packages are required depending on your current OS. It is possible to override the automatic settings and specify them manually, though

node['unicorn-ng']['packages'] = %w(rubygems)

Configuration

Everything in your unicorn.rb can be maintained using attributes. Consider using the provides LWRPs (see below)

Most importantly, you need to specify the path to your unicorn.rb. If this is not specified, the default recipe will do nothing. You can also specify a working directory, if needed

node['unicorn-ng']['config']['config_file'] = '/var/www/examples.com/config/unicorn.rb'
node['unicorn-ng']['config']['working_directory'] = '/var/www/examples.com'

This section describes the supported attributes, as well as their default settings.

node['unicorn-ng']['config']['worker_processes'] = 1
node['unicorn-ng']['config']['listen'] = 8080
node['unicorn-ng']['config']['backlog'] = nil
node['unicorn-ng']['config']['pid'] = 'tmp/pids/unicorn.pid'
node['unicorn-ng']['config']['timeout'] = 60
node['unicorn-ng']['config']['stderr_path'] = 'log/unicorn.stderr.log'
node['unicorn-ng']['config']['stdout_path'] = 'log/unicorn.stdout.log'
node['unicorn-ng']['config']['preload_app'] = true

# When sent a USR2, Unicorn will suffix its pidfile with .oldbin and
# immediately start loading up a new version of itself (loaded with a new
# version of our app). When this new Unicorn is completely loaded
# it will begin spawning workers. The first worker spawned will check to
# see if an .oldbin pidfile exists. If so, this means we've just booted up
# a new Unicorn and need to tell the old one that it can now die. To do so
# we send it a QUIT.
#
# Using this method we get 0 downtime deploys.
#
# Stolen from: https://github.com/blog/517-unicorn
node['unicorn-ng']['config']['before_fork'] =  <<-EOS
  old_pid = '#{node['unicorn-ng']['config']['pid']}.oldbin'
  if File.exists?(old_pid) and server.pid != old_pid
    begin
      Process.kill('QUIT', File.read(old_pid).to_i)
    rescue Errno::ENOENT, Errno::ESRCH
      # someone else did our job for us
    end
  end

  if defined?(ActiveRecord::Base)
    ActiveRecord::Base.connection_handler.clear_all_connections!
  end
EOS

node['unicorn-ng']['config']['after_fork'] =  <<-EOS
  if defined?(ActiveRecord::Base)
    ActiveRecord::Base.connection_handler.verify_active_connections!
  end
EOS

It's also possible to add arbitrary commands to the top of unicorn.rb in case you need something like a special require statement.

node['unicorn-ng']['config']['prescript'] = <<-EOS
  # Load the GELF logging extension
  require 'gelf'
  logger ::GELF::Logger.new('graylog', 1234, 'LAN', {facility: 'unicorn'})
EOS

Furthermore, you can define more advanced settings, if needed (ignored when using systemd)

node['unicorn-ng']['config']['owner'] = 'root'
node['unicorn-ng']['config']['group'] = 'root'
node['unicorn-ng']['config']['mode'] = 00644
node['unicorn-ng']['config']['cookbook'] = 'unicorn-ng'
node['unicorn-ng']['config']['source'] = 'unicorn.rb.erb'
node['unicorn-ng']['config']['variables'] = {}

Service

This cookbook can set up unicorn so it gets properly started at boot time.

Analogue to the configuration, you need to specify the path to your rails application. If this is not specified, the default recipe will do nothing.

node['unicorn-ng']['service']['rails_root'] = '/var/www/example.com'

The following attributes will be set automatically relative to the rails_root, if not specified.

node['unicorn-ng']['service']['config'] = nil
node['unicorn-ng']['service']['bundle_gemfile'] = nil
node['unicorn-ng']['service']['pidfile'] = nil

If you need a different bundler (e.g. a wrapper from rvm), you can specify it here

node['unicorn-ng']['service']['bundle'] = '/usr/local/bin/bundle'

It's also possible (since 0.2.0) to specify a wrapper (like chruby-exec) (disabled by default)

node['unicorn-ng']['service']['wrapper'] = '/usr/local/bin/chruby-exec'
node['unicorn-ng']['service']['wrapper_opts'] = "#{my_ruby_string} --"
node['unicorn-ng']['service']['bundle'] = 'bundle'

In the same way, you can set your custom GEM_HOME path:

node['unicorn-ng']['service']['gem_home'] = '/usr/local/ruby/gems'

Since version 1.2.1, you can specify the --chdir flag for start-stop-daemon, which results in a chdir before the app starts

node['unicorn-ng']['service']['chdir'] = '/home/myapp'

The RAILS_ENV. Set this to 'production' in your production environment

node['unicorn-ng']['service']['environment'] = 'development'

The user unicorn runs at. NOTE: THIS SHOULD BE CHANGED TO AN UNPRIVILEDGED USER

node['unicorn-ng']['service']['user'] = 'root' # CHANGE ME! (e.g. 'www-data')

The locale (set by the initscript)

node['unicorn-ng']['service']['locale'] = 'en_US.UTF-8'

Use systemd (defaults to true on Ubuntu >= 15.04) NOTE: This is recommended on machines with systemd, as it's a way cleaner solution. However, the 'status', 'add-worker' and 'remove-worker' actions are not supported on systemd. Furthermore, systemd restart equals former full-restart, reload is behaving like former restart and the former reload was dropped.

node['unicorn-ng']['service']['systemd'] = false

Since 0.3.0, you can specify the service name. The initscript will be deployed to /etc/init.d/$SERVICENAME, or, when systemd is used, to /etc/systemd/system/$SERVICENAME.service Defaults to 'unicorn'

node['unicorn-ng']['service']['name'] = 'unicorn'

Additional options for the initscript (if required, ignored when using systemd)

node['unicorn-ng']['service']['owner'] = 'root'
node['unicorn-ng']['service']['group'] = 'root'
node['unicorn-ng']['service']['mode'] = 00755
node['unicorn-ng']['service']['cookbook'] = 'unicorn-ng'
node['unicorn-ng']['service']['source'] = 'unicorn.init.erb'
node['unicorn-ng']['service']['variables'] = {}

Recipes

default

Configures unicorn.rb configuration, as well as the unicorn initscript according to the attributes above. Unless you specify at least one of the following attributes, this recipe will do nothing.

node['unicorn-ng']['config']['config_file'] # path to your unicorn.rb (will configure unicorn.rb)
node['unicorn-ng']['service']['rails_root'] # path to your rails_root (will configure the unicorn service)

install

Installs bundler and unicorn using the systems "rubygems". Consider installing a more recent environment using tools like rvm.

Provider

unicorn_ng_config

Configures unicorn.rb. This LWRP uses the given attributes (see above) as defaults, if not specified otherwise.

Example:

unicorn_ng_config '/var/www/example.com/shared/config/unicorn.rb'

Or, a more complex setup:

unicorn_ng_config '/var/www/example.com/shared/config/unicorn.rb' do
  worker_processes 16 if node.chef_environment == 'production'
  worker_processes  6 if node.chef_environment == 'staging'
  worker_processes  2 if node.chef_environment == 'development'

  case node.chef_environment
  when 'production', 'staging'
    # We may be started by root, thus dropping privileges
    user 'deploy'
    working_directory '/var/www/example.com/current'

    # Listen on UNIX domain socket only
    # Shorter backlog for quicker failover when busy
    listen  'unix:tmp/sockets/unicorn.sock'
    backlog 1024

  when 'development'
    listen 8080
  end

  # Kill workers after 30 seconds on production
  timeout (node.chef_environment == 'production' ? 30 : 60)
end

unicorn_ng_service

Deploys and configures unicorn initscript. This LWRP uses the given attributes (see above) as defaults, if not specified otherwise.

Example:

unicorn_ng_service '/var/www/example.com/current'

In case you e.g. want to use the unicorn bundled with your configuration, you may set up a rvm wrapper for bundler

Using fnichol-rvm:

rvm_wrapper 'init' do
  binary 'bundle'
  ruby_string 'ruby-2.0.0@myapp'
end

Or do it manually

rvm wrapper ruby-2.0.0@myapp init bundler

The initscript then uses this bundler to call unicorn with

bundle exec unicorn [ARGS]

Then tell the service to use your custom bundler:

unicorn_ng_service '/var/www/example.com/current' do
  environment   node.chef_environment
  user          'deploy'

  # use a custom bundle-environment
  bundle        '/usr/local/rvm/bin/init_bundle'
end

An example with chruby-exec

unicorn_ng_service '/var/www/example.com/current' do
  wrapper       '/usr/local/bin/chruby-exec'
  wrapper_opts  '1.9.3-p448 --''
  bundle        'bundle'
  environment   node.chef_environment
  user          'deploy'
end

Contributing

Contributions are very welcome!

1. Fork the repository on Github
2. Create a named feature branch (like add_component_x)
3. Write you change
4. Write tests for your change (if applicable)
5. Run the tests, ensuring they all pass
6. Submit a Pull Request using Github

License and Authors

Author: Chris Aumann [email protected]

Contributor: Moshe Bergman [email protected]

License: GPLv3