Cyril Chapellier

👋🏼 I can try to craft a PR, but I'm wondering why we can't use the intermediate queue as per the "lmove" scenario: When dequeuing: ```python result = connection.blpop(queue_keys, timeout)...

👋🏼 If I understand correctly, the problem in this issue is exactly the one the intermediate queue solves, aka "_jobs get lost in between dequeuing and the started registry._" (I...

Hi @AntiSol Allowing myself to ask a question on this issue report: > I think what's probably happening here is that one or more failed jobs has hit their expiry...

👋🏼 Piggy-backing on this because I've seen a somewhat similar behavior: - Calling `failed_job_registry.get_job_ids()` does indeed call `failed_job_registry.cleanup()` ([source](https://github.com/rq/rq/blob/master/rq/registry.py#L152)) so you shouldn't need to call it beforehand (the case of...

Gentle poke @selwin in case you have some insights to share 🙏🏼

> Ideally we shouldn't have a situation where a job ID in the FailedJobRegistry with a future expiration date, but the job hash itself is no longer present. The only...

Thanks for all the precisions. The fact that the score (in the failed registry) is in the future makes the bug weird, as it's probably not RQ removing the job...

Thanks for the pointers. Do you have a TL,DR of the improvements you made (apart from adding the reverse-proxy in there), that would be helpful to bake into a common...

Any chance this could be merged and published ? This is a quick win

Hi @chadicus Could you elaborate on what parts of the 4.x PR you would need help for ? I've got a bit of time but I'm not sure where to...