potiuk opened a new pull request, #76185: URL: https://github.com/apache/airflow/pull/76185
The fork CI monitor posted a "Could not classify or update" item to `#internal-airflow-ci-cd` each time reading a fork run's jobs hit a transient GitHub server error. On 2026-10-31 that produced three posts for one pull request within 75 minutes: - the first HTTP 502; - a repeat after three clean runs, since an item that goes away and comes back is posted again; - another repeat when the error changed from HTTP 502 to a reset HTTP/2 stream, since the item embedded the error text. Failing to post job statuses never blocks reconciling, so a one-off failure needs no alert. Like rate limiting and deferred job statuses, the monitor now counts consecutive failing runs (per pull request here) in the record it carries between runs. A pull request is reported only after three failed runs in a row, about 45 minutes. The item's wording is fixed, so a persisting failure is posted once whatever the error. Each run's error is still in the run log. Tested with `uv run --project dev/breeze pytest dev/breeze/tests/test_fork_ci_reconcile.py` (323 passed) and the pre-commit hooks. --- ##### Was generative AI tooling used to co-author this PR? - [X] Yes — Claude Code (Opus 5.5) Generated-by: Claude Code (Opus 5.5) following [the guidelines](https://github.com/apache/airflow/blob/main/contributing-docs/05_pull_requests.rst#gen-ai-assisted-contributions) 🤖 Generated with [Claude Code](https://claude.com/claude-code) -- 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]
