#celery

any good ways to deal with reliability during worker/instance termination when using a redis broker, or is the answer basically to use rabbitmq?

i'm sometimes getting celery to return tasks when the instance knows termination is coming, but not always, and then sometimes i have the opposite issue where the returned tasks get run too many times by other workers

need help flower in supervisor is not able to track my celeryd

I am new to celery

with celery4 (message protocol v2), how can i modify the args/kwargs in a before_task_publish handler?

previously i used this code which worked just fine: body['args'] = _CelerySAWrapper.wrap_args(body['args']); body['kwargs'] = _CelerySAWrapper.wrap_kwargs(body['kwargs'])

but now the body is a tuple and thus immutable... and while i can modify the task kwargs in place since it's a dict, i can't do that for args since it's also a tuple..

or is there some other (betteR) way to preprocess task arguments?

hrm, i guess i could override apply_async..

also, is there some plugin for celery that adds the email functionality back or do i really need to implement this myself?!

https://github.com/celery/celery/issues/3588 - there's no solution in there regarding how to log tasks where the worker gets killed with SIGKILL, e.g. by the kernel's oom-killer. however, celery 3.x did exactly this. how do i get back this functionality? on_failure apparently does not run in this case.

Hi guys. Are you aware that the http://www.celeryproject.org/ page is happily announcing to the world that you have had zero downloads in the last month? ;-)

Hi! Let's say i want to run 10 tasks, but only at most 4 parallely (at one time). Is the following scenario possible to (easily) achive with celery? 1) run task (it can be in one .sh file, it can be a few commands, I could adjust). If failed, rerun. So, if tasks 1,2,3,4 are running and for example task 1 fails, then i would want to rerun task1 before running next task (task 5)

https://github.com/getsentry/raven-python/issue... - did anyone manage to log WorkerLostErrors with sentry by any chance?