mobuchowski commented on code in PR #43936:
URL: https://github.com/apache/airflow/pull/43936#discussion_r1848943761
##########
providers/src/airflow/providers/dbt/cloud/utils/openlineage.py:
##########
@@ -156,6 +156,8 @@ async def get_artifacts_for_steps(steps, artifacts):
task_id=operator.task_id,
logical_date=_get_logical_date(task_instance),
try_number=_get_try_number(task_instance),
+ queued_dttm=task_instance.queued_dttm,
Review Comment:
>We recently added a feature, when a task gets stuck in queued, then we may
set it back to scheduled and queue it again.
It's not a problem, because it's not going to running state - listeners
never get called in that case.
>Did you have any particular reason that you thought it was needed? (Beyond
a reason you think it wouldn't hurt)
The issue with reschedule is getting duplicate calls to listener, with all
parameters the same, besides queued_datetime. This causes duplicate OL start
events with same run_id, which is against OL spec. We don't have other way to
distinguish those, as far as I see. Adding something new, like reschedule
counter, is not a good idea - we don't even know whether reschedule will be in
3.0, and it won't fix things for 2.8-2.10 users.
--
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]
