SEPURI-SAI-KRISHNA commented on PR #12381: URL: https://github.com/apache/seatunnel/pull/12381#issuecomment-5789122035
That is a much better data point than anything I brought, thanks for going and checking the fix's own validating run. A 60 second `arrived.await` consuming its entire budget, in the run meant to prove 60 seconds is enough, is the part that should decide how this is treated. Agreed on your reading. Two successive widenings, 15s then 60s, and the arrival signal has failed to fire inside the budget at each one. A full minute with no request reaching a same-process loopback `HttpServer` is hard to explain as scheduling delay alone, and a swallowed failure before `arrived.countDown()` would look exactly like this from the outside. On the line numbers you flagged: mine were read off `dev`, so pre-patch, and yours are the file with this PR applied. The offset is exactly the 8 lines this PR inserts for the new constant and its javadoc, which is why 395/397/416/417/423 and 403/405/424/425/431 line up one for one. I should have said "on dev" rather than leaving it to be inferred. Our own CI failures are reported at `:416` for the same reason, since those runs are on `dev` without this PR. No change to my view of the diff itself: it widens rather than weakens, and the behaviour bounds stay at 15s. I agree with merge plus a fast-follow on the root cause rather than waiting for a third widening. -- 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]
