<<Bumping it again, in case it was missed earlier>> On Fri, Feb 15, 2019 at 12:49 PM [email protected] < [email protected]> wrote:
> Hi team, > > We wanted to enable dag_run timeouts on our dags but when we have gone > through the behavior of how dagrun_timeout works, we got to know, it only > works on when below conditions are met. > > 1) dagrun should be scheduled one i.e. not manually created > 2) max_active_runs must be configured > > When it works: > During dagrun creation, if count of existing dagruns equate to > max_active_runs configured and previous runs are running longer than > configured timeout, it will be marked as failed( > https://github.com/apache/airflow/blob/master/airflow/jobs.py#L784) > > How can we achieve below: > > 1) With manually created dag_runs > 2) Enabling timeouts on existing dag_runs without requiring triggering of > new dag_runs > 3) Though dag_run is marked as failed but running task will keep on > running until it reaches terminal state. > > Workaround: > We also explored execution_timeouts at an individual task level in > combination with corresponding trigger rule, this works perfectly for us. > > Regards, > Vardan Gupta >
