albertopinardi commented on issue #26394: URL: https://github.com/apache/airflow/issues/26394#issuecomment-1248595259
Hey! I'm facing the same issue and I have a similar configuration. I'm going to share what I tested so far, but first a bit of background: - I had a perfectly functional 2.2.3 on Kubernetes[EKS 1.21] (installed with the bitnami/airflow Helm Chart). - I upgraded straight to the 2.3.4 doing a `airflow db upgrade` and everything was fine (for the first 10 minutes). - Tasks won't schedule with an unrecognizable pattern (some of them would run manually but when the time comes they won't start or the other way round. Totally random, no clue). This is what I tested so far, without much luck, anyway. - I thought it was something wrong with the metadata store (PostgreSQL in my case) so I cleaned some DAGs history deleting and re-importing them but without success. - Rebooting the scheduler would (sometimes) restart the queued process, but it would eventually start to stall eventually. - I created a new namespace and replicated a fresh installation of a 2.3.4 I tested some DAGs and the problem is still there. I think there is something wrong with this version and i will rollback to a 2.2.5 tomorrow, because i have some production workload that i would like to keep going. If i have some info on this, i would reply here. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
