Selwin Ong

> When the worker finishes a job, could it pull the batch ID and extend the TTL of the batch by some user-specified value? This means as long as the...

> Sounds great, I’m happy to take a stab at it. Please don't hesitate to ask for review in small batches, to make sure that the general direction is correct...

@dino-rodriguez I don't understand how `None` is put on the registry. Mind showing me a snippet of how you retrieve jobs from the `FailedJobRegistry`?

Sorry for the late reply: can you do this? ```python def retrieve_jobs(queue_name: str) -> None: queue = Queue(queue_name, connection=_connection) for job in queue.failed_job_registry.get_job_ids(): _logger.info(job) # This is the job ID...

Sorry I missed your message. > are you aware of a change / bug that would make jobs being deleted before they actually expire, without removing the key from the...

> Other than that, is the removal of a job from the failed registry atomic, or could it be incomplete because a worker was killed during the process (we have...

Are you able to find out whether the missing jobs are all of the same type? Just wondering whether there's a combination of job TTL's or what not that may...

> Are you able to find out whether the missing jobs are all of the same type? Just wondering whether there's a combination of job TTL's or what not that...

Mind providing a test case for this?

I'm ok with adding more callbacks to RQ, whether the callbacks are for individual status changes such as `on_stopped` or a global one `on_status_change` like what @jpsnyder proposed. PR welcome...