SEPURI-SAI-KRISHNA commented on PR #12927: URL: https://github.com/apache/seatunnel/pull/12927#issuecomment-6366167772
Thanks for re-reading the whole diff at the new head. The `kafka-connector-it` result you asked for is in. A correction of mine first. **Correction.** I said `compatible_kafka_connect_json` was out of scope because I had not run it. I have now, and it fails the same way: #12966 and PR #12967. Treat my "I will not assert an outcome" line as superseded. Your issue 2 is now covered in three places: here, #12953 and #12967. **CI on head `dcc48f426`**, run `39815781771`, all 96 jobs read at job level: **84 success, 11 skipped, 1 cancelled, and no failing test anywhere.** - `kafka-connector-it (11)` success, 03:08:08 to 06:03:07; `(17)` success, 03:37:03 to 06:36:30 - `engine-v2-it` 11 and 17 success, so backpressure did not recur - all four `transform-v2-it` and three of four `unit-test` legs success On both legs `KafkaIT` reports `Tests run: 238, Failures: 0, Errors: 0, Skipped: 0`, and the logs show six distinct topics, `test_topic_start_mode_<engine>_avro-tombstone_9` through `_14`, one per engine container. That is the isolation observed, not inferred, and `Skipped: 0` means nothing was quietly disabled. The gate is not satisfied yet, though. The single non-success is `unit-test (11, windows-latest)`, stopped at exactly `timeout-minutes: 120` (17:47:27 to 19:47:37), so the run conclusion is `cancelled` and the apache-side `Build` reads `timed_out`, which `update_build_status.yml` derives from the timeout annotation. `Upload test reports` is `if: failure()`, so that job left no artifact; I read its raw log instead. No test failed. It is the hang you diagnosed in #12640, a third occurrence, node `[localhost]:5804` again. At 18:42:02: ``` .114 5804 All node is lite node, shutdown this cluster .114 5804 is SHUTTING_DOWN .114 5804 Node is already shutting down... Waiting... .130 5805 Terminating forcefully... ``` 5805 then completed in 3024 ms; 5804 then logged `The Node is not ready yet, Node state PASSIVE` 393 times until the kill. Your reverse-order teardown would have saved this run. Rerunning that one job is enough, with no new push, since `update_build_status.yml` re-reads the run's jobs every 15 minutes. **One cross-PR doc point.** `Kafka.md:625` here says "The `protobuf` format has no such guard", and the list above it omits `compatible_kafka_connect_json`. #12953 and #12967 each change that fact and neither touches `Kafka.md`, to keep one problem per PR. If this lands first I will update that sentence in whichever of the two merges after it. -- 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]
