dag.test(): Flush session() before processing Executor Event Buffer

[juditnovak]

---

closes: #59074

The issue have been demonstrated on separate branch (including helper code that is not to be merged): juditnovak#2

The fix have been demonstrated to respond to the issue (including helper code that is not to be merged): juditnovak#1

Root cause

The dag.test() function is using a single session throughout its lifetime. Since data is cached on the session, outdated results are retrieved.

1. LocalExecutor executes workloads via the supervisor
2. at the end of the task execution, the supervisor is sending an call to the Airflow API /execution/task-instances/ endpoint. This call is responsible for updating the database with the final state
3. then the supervisor returns results to the Executor, who will register the results in its Event Buffer
4. NOTE: 1.)-3.) were part of the Executor's heartbeat()/sync() calls.
5. in the particularly intense dag.test() polling workflow the SchedulerJobRunner.process_executor_events() call is invoked right after, rapidly processing contents of the Executor's Event Buffer.
6. however, at this point DB updates from API call from 2.) are not yet available on the dag.test() session. Instead a cached record is returned from an earlier update.

Testing

• Currently I find not tests within the task-sdk targeting dag.test(). I don't know if this may be intentional?
• I find a single Airflow unit test directly relying on the use_executor=True option
  • it is not the target of the tests, but a tool to perform other tests
• In the supplementary material to this PR I'm providing showcase to the fix. However since the issue is related to cache expiration, I'm afraid it would be hard to reproduce with determinisic (i.e. 'non-flaky') tests.

[boring-cyborg]

Congratulations on your first Pull Request and welcome to the Apache Airflow community! If you have any issues or are unsure about any anything please check our Contributors' Guide (https://github.com/apache/airflow/blob/main/contributing-docs/README.rst)
Here are some useful points:

• Pay attention to the quality of your code (ruff, mypy and type annotations). Our prek-hooks will help you with that.
• In case of a new feature add useful documentation (in docstrings or in docs/ directory). Adding a new operator? Check this short guide Consider adding an example DAG that shows how users should use it.
• Consider using Breeze environment for testing locally, it's a heavy docker but it ships with a working Airflow and a lot of integrations.
• Be patient and persistent. It might take some time to get a review or get the final approval from Committers.
• Please follow ASF Code of Conduct for all communication including (but not limited to) comments on Pull Requests, Mailing list and Slack.
• Be sure to read the Airflow Coding style.
• Always keep your Pull Requests rebased, otherwise your build might fail due to changes not related to your commits.
  Apache Airflow is a community-driven project and together we are making it better 🚀.
  In case of doubts contact the developers at:
  Mailing List: dev@airflow.apache.org
  Slack: https://s.apache.org/airflow-slack

[potiuk]

Usually we require unit tests, but dag.test is a long and not unit-testable really, and the dag.test on its own is used to run tests, so it's a bit "who tests the tester".

It would likely be great however (as a follow-up) to add a test to one of our e2e tests to make a complete run of dag_test with executor and without. Not a unit test but still should allow us to actually test the "dag.test()".

I will create an issue for that.

[potiuk]

Ok. old issue from the past with timeout - looks good. Merging.

[boring-cyborg]

Awesome work, congrats on your first merged pull request! You are invited to check our Issue Tracker for additional contributions.

[github-actions]

Backport successfully created: v3-1-test

Status	Branch	Result
✅	v3-1-test