When workers lose connection to the redis broker as happens time to time in the cluster due to restarts/upgrades/etc, they stall consuming and do not self-recover. Looks like this was addressed in Kombu 5.4.0, so we'll upgrade the whole celery stack to fix.
Description
When workers lose connection to the redis broker as happens time to time in the cluster due to restarts/upgrades/etc, they stall consuming and do not self-recover. Looks like this was addressed in Kombu 5.4.0, so we'll upgrade the whole celery stack to fix.