KUMAR-AKASH-M commented on issue #26574: URL: https://github.com/apache/airflow/issues/26574#issuecomment-7030936113
Hi, thanks for the reply. I looked at what is on main now. `airflow jobs check` still only reads the database and checks the heartbeat, and the health endpoint does the same thing in a different shape. So a scheduler that died recently, or one stuck but still heartbeating, still looks alive. The one thing that looked like a real fix to me is the local health file the dag processor is getting in #74794. The processor writes the file itself on every heartbeat, so a stale file means the process is gone, and it needs no database access. Would it make sense to do the same for the scheduler once that lands? I could add the same health file write to the scheduler loop and a small check flag, and leave the triggerer for later. Happy to go smaller instead if you would rather keep this narrow. -- 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]
