DanielLeens commented on PR #12444: URL: https://github.com/apache/seatunnel/pull/12444#issuecomment-5846306493
Thanks for going to the trouble of stress-testing this locally — that kind of independent verification is exactly what a flake fix like this needs, since the failure was always intermittent and easy to "fix" by accident. Your numbers line up well with the theory: the base failures burning the *entire* wait budget (15s, then the full 60s under #12381) is consistent with the latch never being counted down at all, not with the request simply arriving late. And forcing the body-write `IOException` to reproduce it 20/20 on base while this PR's ordering stays at 0/20 is a good way to isolate the actual mechanism rather than just re-running the flaky path and hoping. That's more rigorous than what CI alone gives us here. On #12454: agreed that landing this PR and dropping the overlapping `PayPalClientTest` change from #12454 is the cleaner path. That PR bundles the same test file into a much larger, unrelated PostgreSQL CDC change, so keeping the PayPal fix scoped to this PR (which only touches that one file) avoids the two drifting apart or conflicting later. I'd support #12454's author trimming that hunk once this merges — happy to leave a note there pointing back here if that's useful. This PR itself hasn't changed since my last pass, and I still see it as ready to merge — the current `Build` failure is the same unrelated `unit-test (ubuntu-latest)` engine-server flake I traced down last time, not anything in `connector-http-paypal`. Thanks again for the extra confidence here. -- 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]
