TLDR:

In this post, I will introduce a really fast deployment tool - Mina, and show you how to deploy a rails application which runs on unicorn with nginx, also I’ll show you how to organize your mina deployment tasks.

Note:

All code in this post can be find here.

About mina

Mina is a readlly fast deployer and server automation tool”, this how the team who built it describes it. The concept behind Mina is to connect to a remote server and executes a set of shell command that you define in a local deployment file(config/deploy.rb). One of its outstanding feature is it is really fast because all bash instructions will be performed in one SSH connection.

Init

To use mina automating deployment, you need to get the following things ready.

  1. a remote sever
  2. create a user for deployment (e.g. deployer) and add this user into sudoer list.
  3. generate a ssh key pair, add the generated public key into GitHub.
  4. create a deployment target fold on the remove server(e.g. ‘/var/www/example.com’)

once you got these things done, run mina init in your project directory, this will generate a deployment file - config/deploy.rb, then set the server address, deployment user, deployment target and other settings in deployment file, like the following:

set :user, 'deployer'
set :domain, ENV['on'] == 'prod' ? '<prod ip>' : '<qa ip>'
set :deploy_to, '/var/www/example.com'
set :repository, 'git@github.com:your_company/sample.git'
set :branch, 'master'

Setup

Then run mina setup, this will create deployment folders, which looks like this:

/var/www/example.com/     # The deploy_to path
|-  releases/              # Holds releases, one subdir per release
|   |- 1/
|   |- 2/
|   |- 3/
|   '- ...
|-  shared/                # Holds files shared between releases
|   |- logs/               # Log files are usually stored here
|   `- ...
'-  current/               # A symlink to the current release in releases/

Provision

It is very common to setup a new server and deploy application on it, it will be good if we can automating this process, here comes the provision task:

    task :provision do
      # add nginx repo
      invoke :'nginx:add_repo'

      queue  "sudo yum install -y git gcc gcc-c++* make openssl-devel mysql-devel curl-devel nginx sendmail-cf ImageMagick"

      #install rbenv
      queue  "source ~/.bash_profile"
      queue  "#{exists('rbenv')} || git clone https://github.com/sstephenson/rbenv.git ~/.rbenv"
      queue  "#{exists('rbenv')} || git clone https://github.com/sstephenson/ruby-build.git ~/.rbenv/plugins/ruby-build"
      queue  "#{exists('rbenv')} || echo 'export PATH=\"$HOME/.rbenv/bin:$PATH\"' >> ~/.bash_profile && source ~/.bash_profile"

      #install ruby
      queue  "#{ruby_exists} || RUBY_CONFIGURE_OPTS=--with-openssl-dir=/usr/bin rbenv install #{ruby_version}"

      #install bundle
      queue  "#{ruby_exists} || rbenv local 2.0.0-p247"
      queue  "#{exists('gem')} || gem install bundle --no-ri --no-rdoc"

      #set up deploy to
      queue "sudo mkdir -p #{deploy_to}"
      queue "sudo chown -R #{user} #{deploy_to}"

    end

    #helper method
    def ruby_exists
    "rbenv versions | grep #{ruby_version} >/dev/null 2>&1"
    end

    def exists cmd
      "command -v #{cmd}>/dev/null 2>&1"
    end

to be able run this taks multi times, I create some helper method detecting whether an executable exists or not. e.g. ruby_exists, exits('gem') these helper method will return if the executable exits, otherwise, it will run next command to install the executable.

With this task, you can get a server ready for deployment in seveural minutes.

Deploy

Once the server is provisioned, you can deploy your application with mina deploy, here is a typical deploy task:

    desc "Deploys the current version to the server."
    task :deploy => :environment do
      deploy do
        invoke :'git:clone'   #clone code from github
        invoke :'deploy:link_shared_paths' #linking shared file with latest file we just cloned
        invoke :'bundle:install' #install bundle
        invoke :'rails:db_migrate' #run database migration
        invoke :'rails:assets_precompile' #compile assets
        invoke :'unicorn_and_nginx' #setup nginx and unicorn config
        to :launch do
          queue '/etc/init.d/unicorn_myapp.sh reload' #reload unicorn after deployment succeed
        end
      end
    end

Unicorn and nginx

to run our application on unicorn and nginx, we need to create our own unicorn and nginx configuration file and start nginx and unicorn with our configuration. here is the task:

    desc "Setup unicorn and nginx config"
    task :unicorn_and_nginx do
      queue! "#{file_exists('/etc/nginx/nginx.conf.save')} || sudo mv /etc/nginx/nginx.conf /etc/nginx/nginx.conf.save"

      queue! "#{file_exists('/etc/nginx/nginx.conf')} || sudo ln -nfs #{deploy_to}/current/config/nginx.conf /etc/nginx/nginx.conf"

      queue! "#{file_exists('/etc/init.d/unicorn_avalon.sh')} || sudo ln -nfs #{deploy_to}/current/scripts/unicorn.sh /etc/init.d/unicorn_myapp.sh"
    end

Conclusion

With these tasks: mina init, mina provision, mina deploy, mina helps you deploying easily with less mistake, have fun with mina!

Exposing event stream from monolith

# 背景随着微服务概念的流行,微服务得到到越来越多的企业的青睐,开发团队开始构建自己的微服务,笔者了解到的企业的微服务之路大致如下:#### 始于单体架构(Monolith)大多巨型应用一开始都是从一个很小的应用逐渐成长起来的,随着业务的增长,更多的功能被加入进来,几年下来...… Continue reading

墨尔本,新生活

Published on September 04, 2015