potiuk opened a new pull request, #76150: URL: https://github.com/apache/airflow/pull/76150
When a run's test jobs all pull the multi-GB CI image at once, GitHub's artifact storage can answer `503 Egress is over the account limit` for minutes. The stash restore's default retries (3 attempts, 5s base) gave up within about 20 seconds: in [this job](https://github.com/apache/airflow/actions/runs/41093474257/job/124205069483) all three attempts hit the same 503 and the job failed after 22 seconds. On GitHub-hosted runners the `stash-restore` wrapper now passes 8 attempts with a 20s base delay to `apache/infrastructure-actions/stash/restore`. Its backoff is exponential with full jitter and capped at 60s per wait, so a job keeps retrying for up to about 6 minutes, and jobs that fail together don't retry together. Downloads that succeed first time are unaffected; the RunsOn S3 path is unchanged. --- ##### 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]
