hussein-awala commented on issue #32091: URL: https://github.com/apache/airflow/issues/32091#issuecomment-1604772168
> It seems that as triggers fire, the link between the trigger row and the associated task_instance for the trigger is removed before the trigger row is removed. This leaves a small amount of time where the trigger exists without an associated task_instance. The database updates are performed in a synchronous loop inside the triggerer, so with one triggerer, this is not a problem. However, it can be a problem with more than one triggerer. It should not happen even if you have multiple triggerers, because each one take a part of the unassigned triggers (base on capacity arguments) and locks the rows in the DB until update them to add its ID. Also I don't believe that we delete the task_instance before delete the trigger since we load them from a dequeue, and you get this exception when you read from it. I think your problem is with line: https://github.com/apache/airflow/blob/a1ba15570219d2fe77466367e84fafa92cbdb24e/airflow/jobs/triggerer_job_runner.py#L685 where your custom trigger doesn't set the task_instance parameter, and keep it None. -- 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: commits-unsubscr...@airflow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org