How to get Capistrano 3 to use RVM ruby?
Asked Answered
Y

1

5

Gemfile:

gem 'capistrano', '~> 3.0.0'
gem 'capistrano-rails'
gem 'capistrano-bundler'
gem 'capistrano-rvm'
gem 'capistrano3-puma'

Deploy.rb:

set :rvm_type, :user
set :rvm_ruby_version, '2.1.1'
set :default_env, { rvm_bin_path: '~/.rvm/bin' }

Production.rb

namespace :rails do
  desc "Open the rails console on primary app server"
  task :console do
    on roles(:app), primary: true do
      execute_interactively "#{current_path}/script/rails console RAILS_ENV=production"
    end
  end

  def execute_interactively(command)
    cmd = "ssh -l deploy 255.255.255.255 -p 21 -t 'cd #{deploy_to}/current && #{command}'"
    info "Connecting to 255.255.255.255"
    exec cmd
  end
end

Capfile:

require 'capistrano/setup'
require 'capistrano/deploy'
require 'capistrano/rvm'
require 'capistrano/bundler'
require 'capistrano/rails/assets'
require 'capistrano/rails/migrations'
require 'capistrano/puma'
require 'whenever/capistrano'
Dir.glob('lib/capistrano/tasks/*.cap').each { |r| import r }

When I run cap production rvm:check the output is:

rvm 1.25.19 (master) by Wayne E. Seguin <[email protected]>, Michal Papis <[email protected]> [https://rvm.io/]
system
ruby 2.1.1p76 (2014-02-24 revision 45161) [x86_64-linux]

Should it not be using user instead of system because I've specified the rvm_type?

When I run cap production rails:console then I get:

INFO Connecting to 255.255.255.255
/usr/bin/env: ruby: No such file or directory
Connection to 255.255.255.255 closed.
Yellowthroat answered 19/3, 2014 at 12:4 Comment(0)
N
9

I don't think you understand how capistrano-rvm works.

Here's the relevant code

capistrano-rvm works by figuring out the location of your RVM installation, and then prefixing your capistrano commands with the relevant mapped commands. The command-map is part of SSHKit.

Capistrano-rvm by default maps gem rake ruby bundle to an rvm-prefixed version. What this means is that whenever capistrano encounters a command, e.g. in the form of execute :bundle it will replace it with, e.g. ~/.rvm/bin/rvm 2.1.1 do bundle

You've completely side-stepped that entire mechanism in your contrived execute_interactively command that specifies the command in-line. In fact, you've completely side-stepped the beauty of capistrano as a whole by setting up your own SSH session!

Also the whole set :default_env, { rvm_bin_path: '~/.rvm/bin' } is not at all needed, that's the reason you're using the capistrano-rvm gem.


As for why cap production rvm:check is displaying system is because the term system is overloaded in this case. What you think it means in this context is "what kind of rvm setup scheme is this, a /usr/local/rvm installation or a ~/.rvm installation"

What it actually means is from this code, i.e. it checks the ruby version that's listed in RVM as the current ruby, which the default installation of ruby on a server is known as the system ruby

Narbonne answered 19/3, 2014 at 18:50 Comment(2)
Thanks, this answer made me go and investigate Cap3 a bit more. I've gotten to the point where my commands all work, but the console switches to inspect mode (gist.github.com/pawel2105/9706550)Yellowthroat
"the relevant code" was a lifesaver! If you take the path of install RVM in your own location rather than a default, you can tell the rvm capistrano plugin about it by putting this in your deploy.rb file: set :rvm_custom_path, '/my/custom/path/to/rvm'Pommard

© 2022 - 2024 — McMap. All rights reserved.