jscheffl commented on PR #45958: URL: https://github.com/apache/airflow/pull/45958#issuecomment-2637951431
> > Then the worker will go to maintenance pending if there are running jobs, and maintenance mode if all jobs have finished. > > The diagram suggest changing mode depends only on the count of jobs. Is there a way for cluster admin to force entering to mantanince mode? (Force kill all existed jobs)? I think a force maintenance still could be implemented. If somebody wants/needs this. The intend of the current implementation is a graceful drain of running jobs. The "pending" stzate is the transition, same like if you send a SIGINT to a Celery worker, then also the worker does not pull new jobs (stps consuming from queue) but will attempt to complete all jobs and then terminate. Yeah, if there is a urgent demand (that is how I do during testing to be faster) I check the jobs list page and then find the task in execution and mark as failed/success. Not a single click solution but basically a manual workaround. Not often used by me, just in testing :-D -- 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]
