li3zhi4 commented on PR #11633: URL: https://github.com/apache/seatunnel/pull/11633#issuecomment-5292955547
@DanielLeens — following up on the current CI state for head `0dbea3718` after multiple reruns. The remaining red job is `kafka-connector-it (11, ubuntu-latest)`, and it has now failed **three consecutive reruns on the exact same failure signature**, while our feature test (`KafkaJsonDefaultValueIT`) has passed every single run. **Failure signature (identical across all three runs, annotation line numbers given):** ``` Run 1 (job 94330487649): 423547/419198/414894/414603/412919 + 276989/276300 (spark) Run 2 (job 94415607650): 423600/419254/414856/414613/412976 + 277444/276753 (spark) Run 3 (job 94668401633): 423348/423204/418821/414496/414249 + 276967/276226 (spark) ``` Each run shows 6× `SeaTunnel job executed failed` + 2× `Run SeaTunnel on spark failed` + `Process completed with exit code 1`. The failing test is the **upstream `KafkaIT.testKafkaToKafkaExactlyOnceOnStreaming`** (line 2091 → `awaitStreamingPipelineReady:1327` → `ConditionTimeout`), whose root cause is the same `UnknownTopicOrPartitionException` in `KafkaSourceSplitEnumerator.getTopicInfo` (line 383) that originally hit `KafkaJsonDefaultValueIT` — the topic-readiness race we already fixed for our test via the warm-up (produce+consume a throwaway record, then `deleteRecords`). **Key facts:** - `KafkaIT` is an upstream test class; this PR does not touch it (`git diff 73989d5fb..0dbea3718` = only `KafkaJsonDefaultValueIT.java`). - `KafkaJsonDefaultValueIT` has passed in every CI run since the warm-up fix (locally 1/1 and on all four CI attempts), so the fix is effective. - Notably `kafka-connector-it (8, ubuntu-latest)` **passed** on the latest rerun — the race reproduces only on the JDK 11 runner so far. **Suggestion:** the same warm-up pattern (force metadata propagation before job submission) applies to the upstream `KafkaIT` too. I'd be happy to prepare a follow-up PR adding a shared warm-up helper for the Kafka e2e base so both tests are robust. Alternatively, if this job can be re-run by a maintainer or the failure is acceptable as an upstream flake, the current PR is otherwise ready. No code changes on this PR's side are needed for the remaining failures. -- 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]
