Rangsh commented on PR #12081: URL: https://github.com/apache/seatunnel/pull/12081#issuecomment-5595441598
@SEZ9 @DanielLeens thanks for the F1–F8 status map against `5b691c615` — pushed `64b0ec85a` for the remaining open items, and below are the commit/hunk pointers you asked for so you can verify against the diff without another status-only summary. ### Pointers for already-landed F2 / F6 / F5 / F7 | Item | Where it landed | | --- | --- | | **F2** `WALWorkHandler` catch-`Exception` + guarded `executeResponse()` | commit **`26ce5cb26`** — `WALWorkHandler.java` (`walEvent` catch widening; `executeResponse` catches `Exception`) | | **F6** shared batch deadline | commit **`26ce5cb26`** — `IMapFileStorage.batchQueryExecuteFailsStatus`; coverage in **`f08bdf0ac`** (`IMapFileStorageBatchDeadlineTest`) | | **F5** Mockito in module POM | commit **`26ce5cb26`** — `imap-storage-file/pom.xml` explicit `mockito-junit-jupiter` test dependency (kept through current head) | | **F7** method-level Javadoc on timed/untimed `get` | commit **`26ce5cb26`**, tightened wording in **`64b0ec85a`** — `RequestFuture.get()` / `get(timeout, unit)` | ### F6 deadline clamping (confirmed) In `batchQueryExecuteFailsStatus` the deadline is computed **once** before the loop. Remaining time is now explicitly `Math.max(0L, deadlineNanos - System.nanoTime())`; when remaining is `0`, the wait is skipped (never passed a negative timeout into `get`). See `64b0ec85a` / `IMapFileStorage.java`. ### F4 — torn trailing record (addressed without truncate-reopen) Reset/reopen via `fs.create(wal.txt)` would truncate the fixed WAL path, so that is intentionally not done. Instead: - Comment on `WALWorkHandler` now points at `DefaultReader`’s incomplete-frame stop (`startIndex + dataLength > datas.length` → `break`). - Added **`DefaultReaderTornTrailingRecordTest`**: one complete frame + a torn trailer → reader returns only the complete record. That is the recoverability justification for keeping the existing writer after a failed append. ### F8 — timeout logging - **`queryExecuteStatus`**: `TimeoutException` → single-line **WARN** with `requestId` + elapsed/limit ms; stack at **DEBUG**. - Batch path: same WARN+DEBUG pattern (was already WARN-only; now includes elapsed). ### CI note on prior head Fork CI for `5b691c615` (`34175708746`): unit-test / engine-v2-it / Code style **green**. Failures were `all-connectors-it-4` and `all-connectors-it-6` on **JDK 8** only (JDK 11 counterparts passed) — outside this module’s change surface. New head should re-trigger CI; happy to dig into those ITs if they flake again on this push. Happy to do a final pass once you’ve checked `64b0ec85a` against F4/F5/F6/F7/F8. -- 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]
