capistrano-deploy_into_docker

Mini support task to deploy app into docker.

Installation

Add in your Gemfile;

gem 'capistrano-deploy_into_docker', '>= 0.2.1', group: :development

HowTo: deploy your rails application into docker

1) Require task on Capfile

Add following like to your Capfile;

require 'capistrano/deploy_into_docker'

This just add deploy_into_docker:commit task.

2) Create new deploy setting for docker

You need to set :sshkit_backend to SSHKit::Backend::Docker.

Here is an example for config/deploy/docker.rb;

set :sshkit_backend, SSHKit::Backend::Docker
set :stage, :production
set :branch, 'production'
set :deploy_to, '/app'
fetch(:default_env).merge!(rails_env: :production, RAILS_SERVE_STATIC_FILES: 1,
			   SECRET_KEY_BASE: 'dummy', DEVISE_SECRET_KEY: 'dummy',
			   DATABASE_URL: 'mysql2://0/must-be-overrided')

set :linked_dirs, %w{log tmp/sockets public/system}
Rake::Task["deploy:set_linked_dirs"].clear # Save assets on each releases dir (not under shared)

server docker: {
    image: 'sugi/rails-base',
    commit: 'myapp-web',
  }, user: 'rails:rails', roles: %w{web app}

Server definition as Docker host

You need to set docker environment as host infomation hash. The hash requires :image (or :container) key to run.

server docker: {image: 'ruby:latest'}

If you set :commit key, run "docker commit" after deploy automatically.

server docker: {
    image: 'sugi/rails-base:latest',
    commit: 'new-image-name:tag',
  }, user: 'rails:rails', roles: %w{web app}

In addtion, you can add any options for "docker run". For example;

server docker: {
    image: 'ruby:latest',
    commit: 'new-image-name:tag',
    volume: ['/storage/tmp:/tmp', '/storage/home:/home'],
    network: 'my-net',
    dns: '8.8.8.8',
    dns_search: 'example.com',
    cap_add: ['SYS_NICE', 'SYS_RESOURCE'],
  }, user: 'nobody:nogroup', roles: %w{web app}

If you want to build multiple images, define servers for each role. For example;

server docker: {
  image: 'sugi/rails-base:2.3',
  commit: 'myapp-web',
  env: {
    APP_SERVER_CMD: "exec bin/rails server puma"
  }
}, user: 'rails:rails', roles: :web

server docker: {
  image: 'sugi/rails-base:2.3',
  commit: 'myapp-job',
  env: {
    APP_SERVER_CMD: "exec bin/delayed_job run"
  }
}, user: 'rails:rails', roles: :app

Tips

Stop to run some tasks only in docker deploy stage

Use Rake::Task["target:name"].clear, for example;

Rake::Task["passenger:restart"].clear

You can stop to run any target by calling clear method in deploy/stage-name.rb.

Deploy as handy development test

If you run docker on localhost, mount current source dir as docker volume to test purpose (Note: I do NOT suggest this tric to build production environment).

For example;

set :sshkit_backend, SSHKit::Backend::Docker
set :stage, :development
set :branch, 'master'
set :deploy_to, '/app'
fetch(:default_env).merge!(rails_env: :development,
   SECRET_KEY_BASE: 'dummy', DEVISE_SECRET_KEY: 'dummy',
   DATABASE_URL: 'mysql2://0/must-be-overrided'
   )
set :bundle_without, 'test'
set :repo_url, '/src'        # <--------- Use source from mounted volume.

fetch(:linked_dirs, []).clear
fetch(:linked_files, []).clear
Rake::Task["deploy:set_linked_dirs"].clear

server docker: {
  image: 'nvc-base',
  commit: 'nvc-test',
  volume: "#{Dir.pwd}:/src",  # <---------- Mount current source as docker volume.
}, user: 'rails:rails', roles: %w{web app}

Copyright

Author: Tatsuki Sugiura

Files are distributed under MIT License.