all_done trigger rule goes upstream_failed when mapped inputs are all skipped #51320
Description
Apache Airflow version
3.0.1
If "Other Airflow 2 version" selected, which one?
No response
What happened?
A DAG is being marked failed despite all tasks that should be skipped. This only happens when trigger_rule is all_done
What you think should happen instead?
Mapped task should be skipped, not upstream failed.
How to reproduce
This DAG has all tasks skip in Airflow 2, so the DAG state is success. In Airflow 3, it has one task upstream_failed and therefore the DAG as a whole is marked failed.
from airflow.exceptions import AirflowSkipException
from airflow.decorators import dag, task
@dag(
start_date=None,
schedule=None,
)
def upstream_skipper():
@task()
def skip(**context) -> int:
raise AirflowSkipException()
@task(trigger_rule="all_done")
def consume_skip_all_done(value) -> None:
"""Fails as upstream skipped"""
print(value)
@task()
def consume_skip_all_success(value) -> None:
"""Skipped"""
print(value)
consume_skip_all_success.expand(value=skip())
consume_skip_all_done.expand(value=skip())
upstream_skipper()
Operating System
Debian GNU/Linux 12 (bookworm)
Versions of Apache Airflow Providers
No response
Deployment
Astronomer
Deployment details
No response
Anything else?
No response
Are you willing to submit PR?
- Yes I am willing to submit a PR!
Code of Conduct
- I agree to follow this project's Code of Conduct