ColtenOuO commented on code in PR #71767:
URL: https://github.com/apache/airflow/pull/71767#discussion_r3807710607
##########
airflow-core/src/airflow/serialization/definitions/dag.py:
##########
@@ -786,6 +786,19 @@ def _process_dagrun_deadline_alerts(
"deadline_alerts.deadline_created",
tags=prune_dict({"dag_id": self.dag_id, "team_name":
team_name}),
)
+ else:
+ required_dagrun_column = {
+ SerializedReferenceModels.DagRunLogicalDateDeadline:
"logical_date",
+ SerializedReferenceModels.DagRunQueuedAtDeadline:
"queued_at",
+ }.get(type(deserialized_deadline_alert.reference))
+ log.warning(
+ "skipping deadline alert because the deadline
reference evaluated to None",
+ dag_id=self.dag_id,
+ run_id=orm_dagrun.run_id,
+ deadline_alert_id=deadline_alert.id,
+
reference_type=deserialized_deadline_alert.reference.reference_name,
+ required_dagrun_column=required_dagrun_column,
+ )
Review Comment:
This `else` branch fires for every `SerializedReferenceModels.TYPES.DAGRUN`
reference type, not just `DagRunLogicalDateDeadline`/`DagRunQueuedAtDeadline` —
`TYPES.DAGRUN` also includes `AverageRuntimeDeadline`, `FixedDatetimeDeadline`,
and `SerializedCustomReference`.
`AverageRuntimeDeadline._evaluate_with` legitimately returns `None` while
the Dag hasn't accumulated `min_runs` completed runs yet, and it already logs
its own `"Not enough completed runs..."` info message for that case. With this
change, every such call also emits a spurious WARNING here, misleadingly
implying the reference evaluated to `None` because of a null DagRun column —
even though `required_dagrun_column` resolves to `None` for this type (the dict
only maps `DagRunLogicalDateDeadline`/`DagRunQueuedAtDeadline`) and this
reference doesn't read any DagRun column at all. The same false-positive
applies to any `SerializedCustomReference` that legitimately returns `None` for
its own reasons.
```python
with DAG(
dag_id="demo",
schedule=None,
deadline=DeadlineAlert(
reference=DeadlineReference.AVERAGE_RUNTIME(max_runs=10, min_runs=5),
interval=timedelta(minutes=5),
callback=AsyncCallback(empty_callback),
),
):
...
```
Trigger it before 5 successful runs exist, and the api-server log shows both
lines back to back:
```
[info] Not enough completed runs to calculate average runtime for
dag_id=... (found 1, need 5)
[warning] skipping deadline alert because the deadline reference evaluated
to None
reference_type=AverageRuntimeDeadline required_dagrun_column=None
```
The second line is a false positive — it repeats on every trigger until the
Dag accumulates enough run history, so it reads as an ongoing warning-level
problem for something that's actually expected behavior.
Suggest scoping the warning to only the two reference types it's documented
for:
```python
if type(deserialized_deadline_alert.reference) in (
SerializedReferenceModels.DagRunLogicalDateDeadline,
SerializedReferenceModels.DagRunQueuedAtDeadline,
):
log.warning(...)
```
so `AverageRuntimeDeadline` / `FixedDatetimeDeadline` /
`SerializedCustomReference` returning `None` don't get this misleading warning
attached.
--
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]
