Restart application no longer working with Cap 3.1 and Rails 4

Question

The following task was working before we upgraded to Rails 4 and Cap 3.1

desc 'Restart application'
task :restart do
  on roles(:web), in: :sequence, wait: 5 do
    execute :touch, release_path.join('tmp/restart.txt')
  end
end

First of all, I know Cap 3.1 doesn't implicitly call :restart anymore so I added the following:

after :publishing, :restart

However, it fails on attempting to 'touch' the restart.txt file so that Apache will reload the application.

cap aborted!
touch stdout: Nothing written
touch stderr: Nothing written
config/deploy.rb:46:in `block (3 levels) in <top (required)>'
config/deploy.rb:45:in `block (2 levels) in <top (required)>'
Tasks: TOP => deploy:restart
(See full trace by running task with --trace)
The deploy has failed with an error: #<SSHKit::Command::Failed: touch stdout: Nothing written
touch stderr: Nothing written
>

Do I still need a restart? It generally seems okay but I'm wondering if there could be issues that come up by not finding a way to do this.

Answer 1

Had a similar issue, tried to run this command on server and got an error touch: cannot touch 'myappdir/releases/20140416074158/tmp/restart.txt': No such file or directory, so I simply added a line to create a release_path/tmp dir:

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

Answer 2

What's the error message for the failure?

This works for me:

namespace :deploy do

  desc 'Restart application'
  task :restart do
    on roles(:app, :web), in: :sequence, wait: 5 do
      execute :touch, release_path.join('tmp/restart.txt')
    end
  end
end

after 'deploy:publishing', 'deploy:restart'

Answer 3

I've modified the given answer to show the solution that has worked for me.

I've just enter the next lines at the end of the config/deploy.rb file.

namespace :deploy do
  desc 'Restart application'
  task :restart do
    on roles(:web), in: :sequence, wait: 5 do
      execute :mkdir, '-p', "#{ release_path }/tmp"
      execute :touch, release_path.join('tmp/restart.txt')
    end
  end

  after :deploy, "deploy:restart"
  after :rollback, "deploy:restart"
end

Doing that, Passenger will restart automatically after each deploy/rollback.

I think it worth mentioning that you can also restart Passenger manually from your development environment each time you run the next line:

cap production deploy:restart

Answer 4

On my cap I got a similar error on a multi-server deploy. It seems that the problem is that it is relying on your running a previous rails command (to generate the tmp directory) before restarting, such as db:migrate.