li3zhi4 commented on PR #11677: URL: https://github.com/apache/seatunnel/pull/11677#issuecomment-6058745098
Thanks \u2014 the three items are all already implemented on `677216dd4f`, and the requested CI result is now posted above.\n\n1. **`SourceSplitEnumerator` javadoc:** I took the narrowing direction (option (a)). `SourceSplitEnumerator.java:52-56` now states only the engine guarantee: `run()` follows `open()` and reader registration, but restored splits may arrive through `addSplitsBack()` before or after `run()`, and dispatch of such late splits is left to each implementation. The stronger queue-and-dispatch behavior is documented on `IncrementalSourceEnumerator`, not imposed on every connector.\n2. **Pre-`run()` window:** comments are present at both `IncrementalSourceEnumerator.java:77-83` (the trailing `run()` assignment pass) and `:105-111` (the `if (running)` path). They document why a pre-`run()` restored split is queued and then dispatched once readers are registered, and why an already-running enumerator immediately retries assignment. `shouldAssignSplitsAddedBackBeforeRun ExactlyOnce` also covers the invariant directly; it was retained even though the later comment said the E2E path was sufficient.\n3. **30-second wait:** the injected-failure wait is now `await().atMost(2, TimeUnit.MINUTES)` at `AbstractMysqlCDCITBase.java:227`, aligned with the surrounding budgets.\n4. **Retry/trigger race:** `mysqlcdc_to_mysql_with_sink_failure_recovery.conf:24-27` now uses `job.retry.times = 3` (with the 15-second interval), leaving room for a replayed failing insert to consume one restart before the trigger removal takes effect.\n5. **Rotated/shrunk logs:** `AbstractMysqlCDCITBase.java:275` uses `substring(Math.min(logOffset, serverLogs.length()))`, so a shorter log retries instead of throwing `StringIndexOutOfBoundsException`.\n\nThe real Docker/Zeta run was 1/1 with zero failures and zero skips, and the CI MySQL CDC connector jobs were green on both JDK 8 and 11. The residual Build reds are the same unrelated dev-side families already documented. There is no fu rther code change needed for these five points from my side; please do the final review on `677216dd4f` when convenient.\n -- 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]
