seanghaeli commented on PR #77087: URL: https://github.com/apache/airflow/pull/77087#issuecomment-6965371332
Two things the backoff leaves open, both about how it fits with recovery. The count resets once a trigger has stayed up for two minutes (`TRIGGER_RECOVERY_UPTIME`), so a watcher that crashes every few minutes is always on its first failure and always waits the base 5 seconds. Should the time a trigger has to stay up grow with its delay, or at least match the cap? A trigger waiting out its delay shows as `failing` in the API and in `triggers.failing`, same as one that's running again after a failure. Is that what we want, or should `failing` mean only the ones waiting to be re-created? -- 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]
