potiuk commented on PR #30669: URL: https://github.com/apache/airflow/pull/30669#issuecomment-1528816474
> Hi all, I'm the author of #30653 and just wanted to share results of monkey patching our large airflow instance containing thousands of DAGs with that PR: we've had no more issues and our Sentry alerts for this issue are silent. tada > > Remember that [scheduler_job_runner.py works around this issue](https://github.com/apache/airflow/blob/058b6ebdd863ff9178007f5750cc4cde4d3c463f/airflow/jobs/scheduler_job_runner.py#L646-L648). All I did was make backfill job runner behave the same, as I understood from looking at TaskInstanceKey and traversing years of history that a wider change would be difficult. > > Totally happy to see official airflow get some fix, even if it isn't mine, but thought I'd share results ^ since my PR got closed. > Ah- thanks for the context. I re-opened it then. Maybe it is indeed worth to implement it then (I will take a closer look - because regardless of try_num calculation fix - which will only be possible to implement in 2.7 , the quick-fix to backfill job of yours might be applicable as a patch in 2.6.* -- 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]
