Installing a Rails application

Installing a Rails application on Joyent Shared Accelerators involves three steps:

1. Configuring and uploading your application, which is the focus of this article.
2. Deploying your application using Mongrel (the preferred way) or Lighttpd.
3. And finally, proxying your port to a domain.

First, let's get things setup and configured.

If you intend to execute any part of your Rails application via Ruby scripts (e.g. for Cron based automated tasks etc.), you should make sure that the Ruby scripts (such as dispatch.fcgi) contain a correct shebang line to the Ruby binary, which is /usr/local/bin/ruby. This where Ruby is located on Joyent's Shared Accelerators and without this change you'll get an error message when you try to start your application.

It's better to be safe than sorry and freeze your Gems and not rely on the versions installed on the shared accelerator. Do a checklist of which gems you are using and their versions numbers.

rails --version
gem list

Ok so you checked it out and found mismatched and/or missing gems.

You want to use the version of rails you have installed on your local machine instead of the version provided. On your local machine run the following command in your project directory

rake rails:freeze:gems

This works for missing and incorrect versions as well. The exception is a gem which requires native libraries (like RMagick does) you can freeze them into vendor just like rails. On your local machine run the following command in your vendor/gems directory.

gem unpack GEMNAME

If you're installing a Radiant site, you'll need to use its own Rake task for unpacking/freezing the Radiant gem. On your local machine run this from the root of your app:

rake radiant:freeze:gems

Once you've unpacked your gems, add the following to your config/environment.rb file to pull all of the newly added gems into your load_path

config.load_paths += Dir["#{RAILS_ROOT}/vendor/gems/**"].map do |dir| 
  File.directory?(lib = "#{dir}/lib") ? lib : dir
end

Note: Using Radiant 0.6.5 this code caused server startup to fail with references to localhosting path. Removing this addition to config/environment.rb worked fine.

In case your application needs a database, you have to create one first:

• Creating a MySQL database
• Creating a PostgreSQL database
• Backing up and Restoring a MySQL database

Some people consider it a good practice to have a separate MySQL user created for their application(s), rather than using the master login. You can learn how to do that here:

• Managing MySQL users

Now you need to choose a good location for your Rails application. If you only need a single application running on a domain, the easiest approach is to use the existing web directory, so that the various app, config, db etc. directories are stored right under the default web directory, and the contents of the Rails public directory goes into the existing web/public directory. Having only one public directory simplifies deployment, and staging a localhosted testing version too.

Contrary to traditional web hosting technologies (e.g. PHP) and old legacy Rails deployment techniques (e.g. using mod_fastcgi to serve your application on Apache), your application does not have to be uploaded into your public directories (i.e. anywhere under your Document root directory) since Mongrel (and Lighttpd) will be proxied to from Apache. In fact, uploading your entire Rails application into your document root directory is considered a bad practice since it could potentially expose some sensitive data of your Rails code or configuration.

If you intend to run more than one Rails application per domain (e.g. you want to have two different subdomains proxying to two different applications), you may want to choose a different location. For instance, you might want to create a single rails or sites directory in your main account home, and have each Rails application under that in it's own directory (e.g. ~/sites/typo, ~/sites/myapp, etc). Keep in mind that you’ll need to use .htaccess to proxy the sites, and managing ~/web/public/ and the app’s public/ folders becomes more involved.

Now it's time to prove to your self that you have everything correctly configured to run on the server. In the example below there is a model called User. If you do not have a model named User then substitute User with the name of one of your models.

cd ~/path/to/rails/application/dir 
script/console production
Loading production environment.
>> User.find 1
=> #<User:0x328bc54  @attributes={}>

Congratulations you have your database correctly configured. And all gems in your config/environment.rb file are correctly frozen or already available on the server. You still might run into some gem issues, but you will see those error messages in your production.log file.

Once your Rails application has been uploaded and configured, you'll now need to to deploy it using either Mongrel (the preferred method) or Lighttpd. Instructions for either approach can be found here:

• Deploying a Rails application with Mongrel
• Deploying a Rails application with Lighttpd