seanmuth commented on issue #68683: URL: https://github.com/apache/airflow/issues/68683#issuecomment-4735950078
Chiming in here from the debugging investigation with the affected customer: (depending on how particular configs are set) this results in the complete stoppage of task scheduling, until the scheduler(s) is restarted, clearing that working set. The issue will trigger again once another adopt loop is run, provided there are enough completed pods to fill the working set. Either way this is a very significant performance regression. I can get a PR up sometime tomorrow most likely based on the proposed solution above. -- 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]
