Running background processes

Overview

You can use Procfiles to ensure that your background jobs run and are monitored. Doing so is as easy as defining them in the root of your application, in a file called Procfile.

Should you wish to use different processes in different environments, you can name your Procfile in the following convention:

Procfile_ENV

For example, to limit specific processes to running only in your development environment, call the file Procfile_development.

How to run background processes

A typical Procfile may look something like this:

worker: rake resque:work QUEUE=*
scheduler: rake resque:scheduler

The commands above would run rake resque:work QUEUE=* and rake resque:scheduler and monitor them. Cloud 66 will attempt to bring processes that go down or crash up again. Processes are also instructed to start when your server is booted. An overall view of your processes is available in your application.

You can reference your application environment variables with a $ before the name. This will be replaced by the actual value in the command executed. As an example, $RAILS_ENV will be evaluated as production if you are in the production environment.

Running processes with unique identifiers

To assign a unique identifier to your process (for example with Sidekiq), use the {{UNIQUE_INT}} notation. For example, your process could look as follows:

worker: bundle exec sidekiq -e production

This integer should be unique across processes, so that multiple processes won't clash, but may not be unique across servers.

Scaling background processes

You can scale your background processes up and down on the process server page. On your application, click Workers in the left-hand nav and scale up a standalone process server for more resources.

Pause and resume background processes

On each process server you can pause and resume your background processes! You will find that in your application under process server and on the top right corner of each server box.

• Pause is scaling down the processes to zero.
• Resume is scaling them back up to the number of processes there were before pause.