Question: How to set process_alive_threshold or a global job timeout on a worker queue type?

[kirkath]

My use case wants to set up a special queue for long_running_tasks. E.g. When i run async_migrations (specifically - data backfills) through special solid_queue backed background job similar to as described here: https://kirshatrov.com/posts/async-migrations

My queue configuration:

- queues: "long_running_task"
      threads: 1
      polling_interval: 5
      processes: 2

I was wondering:

• Can I set the process_alive_threshold per worker queue, e.g. all are default, but this long_running_task's threshold is at 10 minutes? (I do not necesserily want to set it to a higher threshold globally since most of my jobs run in under 30 seconds)
• Can I configure a global timeout for a job, e.g. tasks that have been running for longer than 30 minutes automatically get shut down with a failure?

Thank you

[rosa]

Hey @kirkath! The process_alive_threshold can only be set globally at the moment, but this threshold is not used to kill workers automatically. This is used by the supervisor to delete process records from the database if a process fails to send the heartbeat for more time than that interval, as described here. It doesn't matter how long jobs take to run, the heartbeat is sent in a separate thread.

Can I configure a global timeout for a job, e.g. tasks that have been running for longer than 30 minutes automatically get shut down with a failure?

Unfortunately, we don't have any feature like this yet. You'd have to do something external to Solid Queue. I'll consider it, in any case.

[kirkath]

Thank you for the quick response. I shall figure something out in the meanwhile. Closing.