Steinthor Palsson

This is probably a bug if it doesn't work Should be no difference between AsyncResult here and with normal celery. What's the result of this now and is it different...

Ah I see what you mean, to check for zombie locks? This still returns "pending" for unknown task IDs, right? So we still don't know whether it's queued or not.

There's not really a reliable way to do this that I know. You can use celery signals to set a custom state instead of "pending" when the task is actually...

Singleton requires redis for the locking backend, but redis does not need to be your broker. Set `CELERY_SINGLETON_BACKEND_URL` to a valid redis url. You can then use rabbitmq as a...

Hi This is not supported currently. This library only provides a custom task class (`Singleton`) and the `apply_async` method of that is where all the magic happens. Directly using `send_task()`...

I see, since hard time limit force terminates and doesn't allow for any cleanup. I think you could set `lock_expiry` to match the time limit https://github.com/steinitzu/celery-singleton#lock_expiry The key gets expiry...

Did you find a solution? If still an issue, could you provide a minimal example to reproduce?

Hey @jtkaufman737 What do you mean by pods exactly? The locking is all coordindated centrally in redis. So this shouldn't happen regardless of how many processes you have dispatching jobs...

I would also log the task ID for this. Check if the tasks have the same ID or different IDs. To rule out any issue with celery itself where 2+...

Hey guys, sorry about the delay. And thanks for submitting this @lsaavedr This looks good at a glance, but I would love some tests to go with it. If either...