SEZ9 commented on issue #11769: URL: https://github.com/apache/seatunnel/issues/11769#issuecomment-5358317845
Thanks @goutamadwant, that is exactly the regression target we discussed. A two-table writer test that simulates `08S01`, fails on the stale statement during buffer replay, reconnects the active writer, rebuilds statements, and asserts both buffered rows are written exactly once covers the gap I described for the multi-table + `generate_sink_sql = true` path. The fact that the test fails when the #9942 handling is removed is a good signal that it is actually pinning the behavior we want. I will review #11879. @NaughtyYang, could you please retry the original failing job against current `dev` (or a release containing #9942 and #10793) and report back whether the `08S01 -> reconnect -> statement closed on replay` failure still reproduces? That result decides whether this issue can be closed once #11879 lands, or whether there is still a functional gap beyond missing test coverage. @zhang-arvin, thanks for offering to help. Since @goutamadwant has already opened #11879 for the regression test, please coordinate with them there rather than starting parallel work. If NaughtyYang's retry shows the failure still reproduces on current `dev`, there will be a functional fix to pick up on top of that test. Concrete remaining asks: 1. @NaughtyYang: retry against current `dev` and share the result (logs if it still fails); 2. review of #11879 to confirm it targets the exact multi-table + `generate_sink_sql = true` replay path; 3. keep this issue open until both the retry result and the regression test are in. #11771 stays scoped to the `connection-test-query` pool-validation prevention path as agreed earlier in this thread. <!-- streview-comment:388 --> -- 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]
