Capistrano3 + Rails4 + NGINX + Passenger. After cap deploy restarts the app, we want to warm it up with a ping, so end user never waits

README.md

Most instructions for using Capistrano tell you how to make it restart Phusion Passenger by 'touch'ing the restart.txt file, but this doesn't immediately restart the app - instead the first person to try to use the application will cause it to be restarted, so they will see a delay of at least a few seconds.

This shows how to add a post-deploy task to 'ping' the server, to cause it to restart immediately.

Then add a deploy:ping task to config/deploy.rb and set it to run automatically after deploy:restart. Alternatively you could put it into the deploy:restart task directly.

TODO: Test the response HTTP header of the ping, that it is a 200, and not some other error response (404/500).

deploy.rb

# config/deploy.rb
set :application, 'sampleapp'

#What is an accessible route to hit on the site?
#Note: I have nginx+passenger listening to :8080, our environment has a different web server entrypoint
#Also, our apps get deployed to a sub-uri, with the application_name as its URI 
set :ping_url, "http://localhost:8080/#{fetch(:application)}"

# This is the standard Phusion Passenger restart code. You will probably already
# have something like this (if you have already got Capistrano set up).
namespace :deploy do

  desc 'Restart application'
  task :restart do
    on roles(:app), in: :sequence, wait: 5 do
      # Your restart mechanism here, for example:
      execute "mkdir -p #{release_path.join('tmp')}"
      execute :touch, release_path.join('tmp/restart.txt')
    end
  end

  after :publishing, :restart

  

  desc 'Warm up the application by pinging it, so enduser wont have to wait'
  task :ping do
    on roles(:app), in: :sequence, wait: 5 do
      execute "curl -s -D - #{fetch(:ping_url)} -o /dev/null"
    end
  end

  after :restart, :ping


  after :restart, :clear_cache do
    on roles(:web), in: :groups, limit: 3, wait: 10 do
      # Here we can do anything such as:
      # within release_path do
      #   execute :rake, 'cache:clear'
      # end
    end
  end

end

routes.rb

# config/routes.rb

SampleApp::Application.routes.draw do
  root to: 'application#index'
end

Changes here from @davejamesmiller's version is that I DO NOT use a special /ping url, I want to test ACTUAL work being done, i.e. my homepage will probably crash if the DB isn't connecting, Elasticsearch doesn't connect, other needed configs...

Further fixes I'd like to do is to test the response of the curl. So, that if I get a 404, or a 500, then I kind of want my capistrano deploy to alert me with a failure? Otherwise, deploys typically work just fine, so I'm happy for now with the app being warmed up, hot&ready, like pizza to go.