The GitHub Actions job "Tests (AMD)" on 
airflow.git/k8s-executor-async-pod-creation has failed.
Run started by GitHub user 1fanwang (triggered by 1fanwang).

Head commit for run:
1a6fcb1183ae83f4ed6cd9ee1a2dfa604a357438 / 1fanwang <[email protected]>
Re-queue worker pods on transient 502/503/504 and connection errors

Following review: rather than KPO's in-line generic_api_retry decorator
(its blocking backoff would stall the scheduler loop, and the whole
run_until_complete batch on the concurrent path), broaden the shared
_handle_pod_publish_error so both creation paths re-queue the transient
failures it previously dropped. 502/503/504 join the re-queue set, with
503's Retry-After honored like 429. Connection-level errors (urllib3 on
the sync client, aiohttp on the async) — which previously crashed the
sequential sync loop or failed the task on the concurrent path — are now
re-queued too, via a TRANSIENT_CONNECTION_ERRORS set shared with
generic_api_retry so the two agree on what counts as transient.

Report URL: https://github.com/apache/airflow/actions/runs/28202954762

With regards,
GitHub Actions via GitBox


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to