trickkiste / capistrano-recipes

Clone and deploy. Deployment made incredibly simple with capistrano. Recipes for many popular infrastructure gems.

Capistrano Recipes

Originally borrowed from Railscasts Episode 337

Installation

in your project's root

git submodule add git://github.com/pferdefleisch/capistrano-recipes config/recipes
cd config/recipes
./install # this copies the deploy-example.rb file to config/deploy.rb
cd ../..
capify .

in your deploy.rb

• add your server information
• remove the recipes you don't want

Recipes

I have marked tasks that occur automatically during deployment when requiring the recipe in config/deploy.rb with [automatic]

asset_pipeline

before

[automatic] precompiles your Rails assets locally

after deploy:update

[automatic] rsyncs them to the server after deploy:update hook

deploy:clean

[automatic] removes your local public/assets folder after deployment

backup

variables

backup_postgres_socket_path: "/var/run/postgres"
backup_encryption_password: prompts for encryption password
backup_backup_server_pass: prompts for backup server password
backup_backup_server_host: prompts for backup server's hostname
backup_backup_server_user: prompts for backu server's username
backup_backup_server_path: prompts for path to backup to on backup server
backup_rsync_push_directories: which directories to backup with rsync - default: ["#{shared_path}/system"]
backup_archive_daily_folders: which folders to archive - default: empty array
backup_archive_weekly_folders: which folders to archive weekly - default: ["#{shared_path}/config"]

install

Installs backup gem config files

symlink

[automatic] symlinks backup.yml file from shared/config/backup.yml to current/config/backup.yml

bundler

[automatic] I have had to hack my way around bundler installs on my server before. I am now comfortable using require "bundler/capistrano" in the deploy.rb file but this is still here for special cases

carrierwave

[automatic] This simply symlinks your shared/uploads path to public/uploads

logrotate

Has a setup task to create a logrotate file for current application. It is pretty hard coded at the moment. The template is in templates/logrotate.erb. The current template is pretty robust, I suggest editing the template if you have any specific needs.

logs

tail

tail your production.log file. There is not currently a way to tail other log files.o

htop

If you have htop installed on your remote server, this will show you the interactive htop screen.

monit

Currently supports nginx, postgresql and unicorn

setup

Installs monitrc. You may need to tweak the set daemon 30 setting, this is how often monit runs its checks.

nginx

install

installs nginx from apt nginx/stable repository

setup

sets up unicorn sites_enabled config file or app

controls

[automatic] start, stop, restart

nodejs

install

Installs nodejs on remote server from ppa:chris-lea/node.js apt repo. This makes memory hungry therubyracer unnecessary. Plus you can now run unbelievably fast and fun to write nodejs utilities on your server :P

passenger

Restarts passenger after deploy

postgresql

variables

postgresql_host: default "localhost"
postgresql_user: default the application var
postgresql_password: prompts for postgresql password
postgresql_database: default "#{application}_production"
postgresql_dump_path: default "#{current_path}/tmp"
postgresql_dump_file: default "#{application}_dump.sql"
postgresql_local_dump_path: default File.expand_path("../../../tmp", __FILE__)
postgresql_pid: default "/var/run/postgresql/9.1-main.pid"

install

Install postgresql apt package on remote server

create_database

Prompts for password and creates production database.
Run before postgresql:setup in deploy:setup hook.

setup

creates database.yml based on postgresql:create_database settings.
Run after postgresql:create_database in deploy:setup hook.

console

Opens an interactive postgresql database console connected to remote server

local:download

Download remote database to local tmp/

local:restore

Restores local database from temp file

local:localize

Dump remote database and download it locally
runs remote:dump then local:download

local:sync

Dump remote database, download it locally and restore local database.
runs local:localize then local:restore

remote:dump

Dump remote database

remote:upload

Uploads local sql.gz file to remote server

remote:restore

Restore remote database

remote:sync

Uploads and restores remote database.
runs remote:upload then remote:restore

pry

console

Opens an interactive rails console with the remote server using pry (pry must be installed in your rails application or irb will be used)

rails_config

[automatic] Symlinks rails_config config files from shared to current

rbenv

variables ruby_version: ruby version to install - default 1.9.3-p125
rbenv_bootstrap: which ubuntu version gist to use - default: bootstrap-ubuntu-12-04

install

Installs rbenv and the bundler gem

resque

requires Rakefile tasks moved into project

[automatic] controls stop, start and restart

unicorn

variables

unicorn_pid: path to your unicorn pid file - default: #{current_path}/tmp/pids/unicorn.pid
unicorn_config: path to your unicorn config file - default: "#{shared_path}/config/unicorn.rb"
unicorn_log - default: #{shared_path}/log/unicorn.log
unicorn_workers: number of child processes unicorn should spawn - default 1

setup

creates unicorn config file and moves it into shared/config/unicorn.rb
creates unicorn init.d control script
adds control script as startup script (update-rc.d)

tail

tails remote server current/log/unicorn.log file

controls

[automatic] start, stop and restart

Other Info

First deploy

cap deploy:install
If you have a new Linux instance you can run this to hook into the install tasks in each of the recipes to install important packages and config files into your system

note: make sure you read the recipe files that you have loaded in deploy.rb because you may need to set config variables like username and path name information

cap deploy:setup
This sets up your directory structure for capistrano in whatever you added to set :deploy_to, "/apps/#{application}" in your deploy.rb file

cap deploy:cold
This is run on your first deployment or if your application is not running. It runs deploy:start on completion instead of deploy:restart

cap deploy
This is what will be run on each subsequent deployment

Order of tasks for cap deploy:cold

note: deploy:create_symlink in versions < 2.10 was called deploy:symlink

• deploy:update
  • deploy:updated_code
    • deploy:finialize_update
  • deploy:create_symlink
• deploy:migrate
• deploy:start # empty task, user defined

Order of tasks for cap deploy

• deploy:update
  • deploy:update_code
    • deploy:finalize_update
  • deploy:create_symlink
• deploy:restart # empty task, user defined

This article is also helpful about task order

Directory Structure after running cap deploy:setup

note: this section is taken almost word for word from RailsCasts #133

myapp/releases
myapp/current -> releases/20081019001122
myapp/shared

Other Resources

• List of all Capistrano default tasks
• List of Capistrano variables

Copyright

Copyright © 2012-2013 Square Bracket eU, Aaron Cruz. See LICENSE.txt for details.