Avner Cohen

I do. While majority of the stuff is already migrated to ReplicaSets, not all of it had.. As far as I read, this is still a supported setup as per...

Getting this also here: ``` invalid literal for int() with base 10: '$PORT,hostname:$PORT,hostname:$PORT' ``` kombu/utils/url.py in url_to_parts at line 81 ``` @property def port(self): port = self._hostinfo[1] if port is...

Yeah, I think the code just assumes a very specific and strict connection URL, and the schema (specially in cluster) is different. In my case, it's not redis sentinel, it's...

Running 2.4.0, we are still getting the same exception: `'NoneType' object has no attribute 'drain_events'` From amqp/abstract_channel.py in wait at line 80 ``` while not p.ready: self.connection.drain_events(timeout=timeout) ``` Anyone else...

:( I afraid not. We randomly get it in production, random workers would crash. @matusvalo What's interesting is that your suggested fix is basically to expose the "correct: exception: `amqp.exceptions.NotFound:...

@matusvalo thanks so much for the insights and time. We are getting this during deploy of new code changes or any maintenance that involves large scale stop/start of workers (btw...

@matusvalo Anything we can help or provide information additional info on that?

@matusvalo On our end, we have seen this behavior when starting some 80 celery workers (with anywhere between 2 to 8 concurrency) in parallel. To work around the issue, we...

@oblador Anything on this item ?

The setTimeout seems to effect the behavior of the master. Removing the printMessage() block makes this work as expected. Interesting behaviour, though, I am wondering if this is a real...