SEZ9 commented on PR #12381: URL: https://github.com/apache/seatunnel/pull/12381#issuecomment-5902632869
@zhangshenghang agreed — let's go with the second option and land the `AWAIT_BUDGET_SECONDS` widening on its own. F2 (wall-clock awaits letting rediscovery land after the `second` append) and F5 (`untilAsserted` draining events so the first observation is lost) are pre-existing characteristics of `rediscoversFileAfterInactiveCursorClosed`, not something this diff introduces. One nuance worth stating for the record: the larger ceiling does make the F5 failure mode slower to surface (a masked failure now takes up to 10s instead of 3s to report), so deferring it is a scope decision rather than a no-op — I'm fine with that trade-off given the PR's one-line purpose. Two small asks so the deferral is traceable: 1. Please open the follow-up issue (or two, if you'd rather split ordering vs. diagnostics) and drop the link here before merge, so F2/F5 have a home outside this PR. 2. A one-line pointer to that follow-up in the `AWAIT_BUDGET_SECONDS` javadoc would help the next person who sees this test flake understand that the ceiling is a margin fix and that the ordering/diagnostics gaps are known and tracked. Not blocking if you'd prefer to keep the diff strictly to the constant. With the `PayPalClientTest` changes no longer part of this PR at `b77b3c2cd4`, F1, F3 and F4 no longer apply here, so I'm treating them as closed for this PR. Once the follow-up link is posted I'm good to approve. <!-- streview-comment:1407 --> -- 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]
