DanielLeens commented on PR #11763: URL: https://github.com/apache/seatunnel/pull/11763#issuecomment-5564183965
Following up with the CI diagnosis I owed from my last comment. The head is still `ac19447ef6` (unchanged since 2026-08-14, so no new commit to re-review), but a fresh `Build` run completed on it since then, and I dug into both of its failing jobs directly in the fork run rather than just re-stating the apache-side pointer status. **`transform-v2-it-part-1 (11, ubuntu-latest)`** — 218/219 tests in the module passed. The one failure, `TestFilterRowKindIT.testFilterRowKindMultiTable`, only failed on 1 of its 8 engine-matrix legs (Flink 1.18.0; Zeta, Flink 1.13/1.15, and Spark 2.4/3.3 all passed on the same commit). On that one leg, the job itself failed at the Assert sink with: ``` AssertConnectorException: ErrorCode:[ASSERT-01] ... row num :150 fail rule: AssertFieldRule.AssertRule(ruleType=MAX_ROW, ruleValue=100.0, equalTo=null) ``` That's on table `test.www` in `filter_row_kind_exclude_insert_multi_table.conf` — a table this PR's `FilterRowKind` transform never touches (it's outside `table_match_regex` and has no `table_transform` override, so it goes through the identity-transform passthrough, not `FilterRowKindTransform`). `test.www` should always get exactly 100 rows; getting 150 on one Flink-only run points to a pre-existing row-count issue in the multi-table sink/source layer, not in the `OptionRule` migration this PR makes (which only touches `FilterRowKindTransform`'s own validation and the two tables it actually filters, `test.abc`/`test.xyz` — neither of those over- or under-counted here). I don't see anything in this diff that could produce that symptom. **`doris-connector-it (8, ubuntu-latest)`** — unrelated to both this PR and the above: `DorisErrorIT.testDoris:85 expected: <true> but was: <false>` in the Doris connector suite, which has a known history of flakiness independent of Transform-V2 changes (the JDK 11 leg of the same job passed on the identical commit). Net: still no code-side blocker, and this fresh run doesn't change that — both failures look environmental/pre-existing rather than caused by this diff. @goutamadwant, once you get the `dev` sync in, a rerun should clear the Doris flake; if `testFilterRowKindMultiTable` fails again on the Flink leg specifically, that's worth its own issue against the multi-table sink/assert path rather than something to chase in this PR. -- 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]
