rammmaa opened a new pull request, #73150:
URL: https://github.com/apache/airflow/pull/73150

   ## Problem
   
   The scheduler scans for task instances without a recent heartbeat 
synchronously in its main loop. If that database query is slow, the scheduler 
can miss its own heartbeat and appear unhealthy, while operators have no metric 
or log message identifying this scan as the cause.
   
   ## Fix
   
   - Measure the scan with the 
`scheduler.task_instance_heartbeat_timeout_scan_duration` timer.
   - Log a warning when the scan exceeds the scheduler health-check threshold, 
including when the query ends in a database error.
   - Register the timer in the metrics registry and cover both sides of the 
threshold plus the error path.
   
   ## Tests
   
   - `breeze ci selective-check --commit-ref HEAD`
   - `breeze testing core-tests --run-in-parallel` (API, Always, CLI, Core, 
Other, and Serialization all passed)
   - Breeze Python 3.10: `pytest 
airflow-core/tests/unit/jobs/test_scheduler_job.py -k heartbeat_timeout -q` (21 
passed)
   - `prek run --from-ref upstream/main --stage pre-commit`
   - `prek run --from-ref upstream/main --stage manual --skip 
compile-ui-assets-dev --skip view-skill-eval --skip run-skill-eval-codex`
   - `breeze build-docs` (all 114 documentation packages and spell checks 
passed)
   
   ## Note
   
   This change adds observability only. It does not alter the scan query or 
task-instance heartbeat-timeout behavior.
   
   closes: #32986
   
   ---
   
   ##### Was generative AI tooling used to co-author this PR?
   
   - [X] Yes — Claude Code (Fable 5.1) and Codex (GPT-5)
   
   Generated-by: Claude Code (Fable 5.1) and Codex (GPT-5) following [the 
guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/18_contributor_documentation.rst#use-of-generative-ai).
   
   ---
   
   Drafted-by: Codex (GPT-5) (no human review before posting)


-- 
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]

Reply via email to