Placing processing into another process

You may decide to move the processing to the different process from the main application. For example, your web application will only enqueue background jobs, leaving their performance to a Console application or Windows Service. First of all, let’s overview the reasons for such decision.

Well scenarios

  • Your background processing consumes too much CPU or other resources, and this decreases main application’s performance. So you want to use separate machine for processing background jobs.
  • You have long-running jobs that are constantly aborted (retrying, aborted, retried again and so on) due to regular shutdowns of the main application. So you want to use separate process with increased lifetime (and you can’t use always running mode for your web application).
  • Do you have other suggestions? Please post them in the comment form below.

You can stop processing background jobs in your main application by simply removing the instantiation of the BackgroundJobServer class (if you create it manually) or removing an invocation of the UseHangfireServer method from your OWIN configuration class.

After accomplishing the first step, you need to enable processing in another process, here are some guides:

Same storage requires the same code base

Ensure that all of your Client/Servers use the same job storage and have the same code base. If client enqueues a job based on the SomeClass that is absent in server’s code, the latter will simply throw a performance exception.

If this is a problem, your client may have references only to interfaces, whereas server provide implementations (please see the Using IoC containers chapter).

Doubtful scenarios

  • You don’t want to consume additional Thread Pool threads with background processing – Hangfire Server uses custom, separate and limited thread pool.
  • You are using Web Farm or Web Garden and don’t want to face with synchronization issues – Hangfire Server is Web Garden/Web Farm friendly by default.

Please use Hangfire Forum for long questions or questions with source code.