amoghrajesh opened a new pull request, #76033:
URL: https://github.com/apache/airflow/pull/76033
<!-- 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:
https://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.
For user-facing UI changes, please attach before/after screenshots (or a
short
screen recording) so reviewers can assess the visual impact.
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".
-->
- [ ] Yes (please specify the tool below)
<!--
Generated-by: [Tool Name] following [the
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions)
-->
`SageMakerTrainingOperator` has no `on_kill`. When a task is killed, times
out, or is cleared, the SageMaker training job it submitted keeps running and
keeps billing, with nothing left in Airflow pointing at it. On GPU or
multi-instance training that is expensive, and the orphaned job has to be found
and stopped by hand in the console.
This adds an opt-in `stop_training_job_on_kill` flag that stops the job
instead, matching `stop_job_run_on_kill` on `GlueJobOperator`. It defaults to
`False`, so nothing changes for existing Dags.
The job name is recorded before the submit call rather than after, so a kill
landing in that window still has a name to act on. It is also recorded after
the timestamp rename that `_get_unique_job_name` may apply, so `on_kill` stops
the job that was actually submitted rather than the name in the config.
`SageMakerHook.stop_training_job` swallows `ValidationException`, which is
what SageMaker returns when the job already reached a terminal state or never
existed. That race is expected when a task is killed as its job finishes, and
it should not surface as a traceback from `on_kill`. Any other `ClientError` is
re raised.
This is the first half of durable execution for this operator. Reconnecting
to a running job on retry, instead of submitting a duplicate, is a follow-up
that will port it to `ResumableJobMixin`.
### Testing
Ran a real SageMaker in `eu-west-1` with a KNN training job with
`stop_training_job_on_kill=True`, marking the task failed moved the job to
`Stopping` and then `Stopped`. With the flag off, the same kill left the job
`InProgress` after the
task was gone.
<img width="2478" height="1190" alt="image"
src="https://github.com/user-attachments/assets/1f76924f-7716-4717-b068-4d11ffcab32e";
/>
```shell
aws sagemaker list-training-jobs --region eu-west-1
{
"TrainingJobSummaries": [
{
"TrainingJobName": "airflow-on-kill-test",
"TrainingJobArn":
"arn:aws:sagemaker:eu-west-1:396807896376:training-job/airflow-on-kill-test",
"CreationTime": "2026-10-30T17:54:15.332000+05:30",
"TrainingEndTime": "2026-10-30T17:54:52.582000+05:30",
"LastModifiedTime": "2026-10-30T17:54:52.864000+05:30",
"TrainingJobStatus": "Stopped",
"SecondaryStatus": "Stopped"
}
]
}
```
---
##### Was generative AI tooling used to co-author this PR?
- [X] Yes (please specify the tool below)
Generated-by: Claude Code (Opus 5) 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`, in
[airflow-core/newsfragments](https://github.com/apache/airflow/tree/main/airflow-core/newsfragments).
You can add this file in a follow-up commit after the PR is created so you
know the PR number.
--
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]
