Should have had our discussion here, sigh...

Pushed a new branch up for testing (deploy_paths). The basic premise is that when capistrano deploys the daemon it will create a new file called ENVIRONMENT in the daemon's root directory. The file is a YAML hash with two respected keys at the moment:

daemon_root: /path
ruby_path: /path/to/interpreter

If the daemon_root key is present (defaults to the capistrano symlink), the rake tasks will overwrite the release specific path to that value. ruby_path is detected by calling 'which ruby' during deploy, if the key is present it will be prefixed in the monit/god config files where appropriate.

Please let me know what you think so I can update the docs and merge to master.

Thanks for reporting !

Is there a way to override the specific version of Ruby to use prior to deploy? I am concerned that 'which ruby' is dependent on the PATH being set the same for the user that cap uses to deploy, and the potentially different path for the user running the script. e.g. what if cap deploys as root who finds ruby in /bin/ruby, and the script runs as user foouser who has /usr/local/bin prepended to PATH so that users version of ruby would be /usr/local/bin/ruby (as in the case where Ruby Enterprise Edition is being used).

Good call, I'll update the deployment recipe so that you can explicitly set the interpreter path and have it fail over to "which ruby" if not set.

Thanks

Just pushed an update to the deploy_paths branch which has this in, it seems to be working fine for me.

In the deployment, set :ruby_path to the correct location of the interpreter.

As a side note, I made the stop & start commands also use the ruby_path which may or may not work as expected based on the availability of the GEM_HOME environment variable.

Please let me know how this looks to you.

Best