Skip to content
A pattern for deploying Wordpress sites via Capistrano.
Ruby
Find file
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.
config
wp-content
Capfile
README.md

README.md

wordpress-cap

A pattern for deploying WordPress sites via Capistrano.

REQUIREMENTS

  • your WordPress site in a git repository (configuration details below)
  • Ruby 1.8.7+ (might work on others, not tested)
  • the following Ruby gems: capistrano, erubis, railsless-deploy
  • a server you can SSH into. (This doesn't really work for shared hosting)

INSTALLATION

You'll need to install the Rubygems above:

sudo gem install capistrano
sudo gem install erubis
sudo gem install railsless-deploy

Then, put the contents of this repo into the ROOT of your WordPress site - the config directory, the Capfile, and make sure advanced-cache.php.erb is in your wp-content if you want to configure caching on deployment.

CONFIGURATION

You'll then need to configure your server for deployment. In your /var/www/site directory - on the remote server - create two directories writeable by the user you'll deploy as (as well as the web user): they should be called shared and releases.

The pattern of deployment is the usual Capistrano routine:

  • all the code gets pushed out to a directory called releases/YYYYMMDDHHMMSS
  • various files that should persist between deployments are symlinked from shared into this release directory.
  • Finally, /var/www/site/current is symlinked to the latest release.

This being WordPress, there are a few things we'd like to persist between deployments:

  • the wp-content/uploads directory
  • the site configuration
  • the cache configuration

To do this, there are some assumptions made about the configuration of your site.

The single most important thing you must know:

DO NOT CHECK YOUR wp-config.php INTO YOUR GIT REPOSITORY

Seriously. Why not? Because this file is almost certainly DIFFERENT for every environment. If you're running WordPress on your laptop, wp-config.php will not be the same as live. What you should do is:

  • add wp-config.php to your .gitignore file
  • put the wp-config.php for the live site into /var/www/site/shared/wp-config.php

We'll then symlink the correct wp-config.php into the site on live. One thing that'll make this easier is defining the WP_HOME and WP_SITEURL constants in your config file: this means you don't have defining these in the configuration database.

Secondly, we don't want to lose our uploads folder on every deploy. So, again:

  • add wp-content/uploads to your .gitignore
  • move your uploads folder in its entirety to /var/www/site/shared/uploads , and make sure it's got the right permissions.

Then, on every deploy, we'll symlink uploads - and any new uploads will thus be shared between releases.

CONFIGURING CACHING

If you use caching on your site - which you probably should - it'd be good not to lose our cache settings on every deploy. Most WordPress caching tools use the ABSPATH constant, which disregards symlinks. However, Capistrano can tell us the full release path when it deploys, so we can generate cache configuration as we deploy.

I have had no luck making the popular WP Super Cache work with this. However, I have successfully made Hyper Cache work with this, so that's what we'll do.

Install Hyper-Cache and set it up as normal. Then, add the advanced-cache.php.erb file (included in this repo) to your local wp-content directory. You might want to alter the .erb file to reflect the options you've selected in your own Hyper Cache configuration. Finally, move your wp-content/cache directory in its entirety - with correct permissions - to /var/www/site/shared/cache.

To enable cache configuration, uncomment the two lines at the end of the rakefile that enable the symlink_hyper_cache and write_advanced_cache_template tasks. On each deployment, this will:

  • symlink the cache directory to the latest release
  • write out a new advanced-cache.php file that'll have the correct full release path in for caching

CONFIGURING PRESERVATION OF LEGACY FILES

Deploying with Capistrano means placing your entire site into version control. That might not be appropriate: if you're like me, you've got a nice tidy WordPress site, and then a bucketload of other scripts and stuff from previous sites. Don't worry! You don't need to put that junk into git if you don't want to. Instead, place it all into /var/www/site/shared/legacy, and uncomment the line in config/deploy.rb that enables the symlink_legacy_files task. Then, on every deploy, everything in your legacy directory will be symlinked into your site root - and it'll be just as it was before, except nicely, tidily deployed.

CONFIGURING APACHE

Needless to say, you'll now need to point your Apache DocumentRoot to /var/www/site/current

PERFORMING THE DEPLOYMENT

Once we've got everything configured, make sure all your code is pushed into your remote git repository. Then, just type

cap deploy

in a shell on your local machine, in the WordPress site root, and everything should get pushed out to your server in one fell swoop.


Something went wrong with that request. Please try again.