crabio commented on issue #36998: URL: https://github.com/apache/airflow/issues/36998#issuecomment-2058807918
@paramjeet01 Fully - not We still have slots leak: <img width="1696" alt="image" src="https://github.com/apache/airflow/assets/40871973/82d34707-fe5c-47d1-b1d7-43ab691985ac";> But we found walkaround: 1. set `AIRFLOW__CORE__PARALLELISM` to 512+ to have free slots in the scheduler. Or bigger than sum of slots in all pools that you have 2. try set `AIRFLOW__SCHEDULER__NUM_RUNS ` to reset scheduler some times. We are going to try this option with 4 hour scheduler life time 3. check that you haven't zombie pods in all namespaces that you are using for the Airflow. We have some namespaces, where, we think, Airflow Scheduler has no access to kill pods and we have zombie succeeded tasks -- 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]
