SEZ9 commented on PR #12454: URL: https://github.com/apache/seatunnel/pull/12454#issuecomment-5851843434
@DanielLeens There is now a concrete failing savepoint-restore CI case for the PostgreSQL path discussed above: - Run `36284739243`, job `108524753463` (`updated-modules-integration-test-part-2`, Java 11): https://github.com/SEZ9/seatunnel/actions/runs/36284739243/job/108524753463 - This is current `dev` (`025d08f9bcb0`) plus the OceanBase-only packaging patch in #12489; it contains no PostgreSQL changes. The failing Zeta job's logged connector jars are CDC base, PostgreSQL CDC and JDBC, with the PostgreSQL driver. - `PostgresCDCIT.testPostgresCdcSnapshotOnlyAndCommittedOffsetStartupModes` fails at line 625: after the restored replication slot becomes active, row `id=15` is inserted but remains absent from the sink for 180 seconds. The class reports 37 tests, 0 failures, 1 error, 0 skips. The same shard passed on Java 8. - The Zeta runtime container is `seatunnelhub/openjdk:8u342`. The restored job `7674347993420909938` logs `lsn_proc=lsn_commit=lsn=36202352` (`0/2286770`), then commits through `0/2286940`, while the row assertion still times out. Relevant log timestamps are 01:40:43–01:40:53 UTC on September 27; the test timeout is reported at 01:43:47 UTC. This is the same startup/restore test that this PR extends to cover a quiet WAL boundary, changes committed while stopped, and subsequent live changes. The log does not include the individual INSERT/BEGIN LSN, so I cannot yet claim that DBZ-6204 is definitively the cause of this particular failure. Sharing the actual runtime case as additional review evidence; no extra patch has been mixed into #12489. -- 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]
