YoannAbriel opened a new pull request, #62878:
URL: https://github.com/apache/airflow/pull/62878
## Problem
On multi-scheduler Airflow deployments (e.g. AWS MWAA with 4+ schedulers),
tasks intermittently fail in bulk with no apparent cause, but succeed on manual
retry. The root issue is in `_executable_task_instances_to_queued`: when the
scheduler checks task concurrency limits and cannot find the serialized DAG, it
immediately bulk-fails **all** SCHEDULED task instances for that DAG via a raw
SQL UPDATE. This is an overly aggressive response to what is often a transient
race condition during DAG file parsing or serialization refresh cycles.
## Root Cause
In `scheduler_job_runner.py`, the `_executable_task_instances_to_queued`
method loads the serialized DAG only when checking per-task or per-dagrun
concurrency limits. If `scheduler_dag_bag.get_dag_for_run()` returns `None`
(serialized DAG transiently absent), the code executed a bulk `UPDATE
task_instance SET state='failed'` for all SCHEDULED tasks of that DAG — instead
of treating it as a transient miss.
## Fix
Replace the bulk-fail with a graceful skip: when the serialized DAG is not
found, log a `WARNING` (down from `ERROR`), add the `dag_id` to `starved_dags`
so the rest of its tasks are also skipped in this iteration, and let the
scheduler retry naturally on the next heartbeat.
**Changes:**
- `airflow-core/src/airflow/jobs/scheduler_job_runner.py`: removed the bulk
`UPDATE … SET state=FAILED` block; replaced with `starved_dags.add(dag_id)` +
warning log.
- `airflow-core/tests/unit/jobs/test_scheduler_job.py`:
- Renamed `test_queued_task_instances_fails_with_missing_dag` →
`test_queued_task_instances_skips_with_missing_dag` and updated assertions to
expect `SCHEDULED` state (not `FAILED`).
- Added new regression test
`test_missing_serialized_dag_does_not_bulk_fail_tasks` (based on the reproducer
from the issue) that explicitly asserts tasks remain `SCHEDULED` when the
serialized DAG is transiently missing.
Both new/updated tests pass. The full `test_executable_task_instances` suite
(24 tests) passes with no regressions.
Closes: #62050
---
<!-- SPDX-License-Identifier: Apache-2.0
https://www.apache.org/licenses/LICENSE-2.0 -->
<!--
Thank you for contributing!
Please provide above a brief description of the changes made in this pull
request.
Write a good git commit message following this guide:
http://chris.beams.io/posts/git-commit/
Please make sure that your code changes are covered with tests.
And in case of new features or big changes remember to adjust the
documentation.
Feel free to ping (in general) for the review if you do not see reaction for
a few days
(72 Hours is the minimum reaction time you can expect from volunteers) - we
sometimes miss notifications.
In case of an existing issue, reference it using one of the following:
* closes: #ISSUE
* related: #ISSUE
-->
---
##### Was generative AI tooling used to co-author this PR?
<!--
If generative AI tooling has been used in the process of authoring this PR,
please
change below checkbox to `[X]` followed by the name of the tool, uncomment
the "Generated-by".
-->
- [X] Yes (Claude Code)
Generated-by: Claude Code following [the
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
---
* Read the **[Pull Request
Guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#pull-request-guidelines)**
for more information. Note: commit author/co-author name and email in commits
become permanently public when merged.
* For fundamental code changes, an Airflow Improvement Proposal
([AIP](https://cwiki.apache.org/confluence/display/AIRFLOW/Airflow+Improvement+Proposals))
is needed.
* When adding dependency, check compliance with the [ASF 3rd Party License
Policy](https://www.apache.org/legal/resolved.html#category-x).
* For significant user-facing changes create newsfragment:
`{pr_number}.significant.rst` or `{issue_number}.significant.rst`, in
[airflow-core/newsfragments](https://github.com/apache/airflow/tree/main/airflow-core/newsfragments).
--
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]
