Hi Abhishek, That sounds very specific. There are a lot of improvements since 1.10. For now, the only advice I can give, please save the logs when this happens again.
Cheers, Fokko Op zo 18 nov. 2018 om 18:36 schreef Abhishek Sinha <abhis...@infoworks.io>: > Hi Fokko, > > This would be difficult to test. I am not sure of the condition when it > occurred. Our DAG runs nightly and this has happened once in the past three > months. > > If this is something that has been identified and fixed in 1.10, I can > make an attempt to upgrade it. But I did not see anything related to this > in the changelog. > > > > > > > Regards, > > Abhishek > > On 18 November 2018 at 8:54:43 AM, Driesprong, Fokko (fo...@driesprong.frl) > wrote: > > Hi Abhishek, > > Do you think it would be difficult to verify if this behavior still exists > in 1.10? Maybe write a dummy dag, and run it using Docker. > > Cheers, Fokko > > Op vr 16 nov. 2018 om 21:45 schreef Abhishek Sinha <abhis...@infoworks.io>: > > > > Airflow Version 1.8.2, Celery and Postgres > > > > Task (and its downstream nodes) got executed even when it's only > immediate > > parent was in upstream failed state. The trigger rule for all nodes was > set > > to "all_success". > > > > Looks like a bug in version 1.8.2. Can someone help me on this? > > > > > > > > > > > > > > > > > > Regards, > > > > Abhishek > > > >
