davidzollo commented on PR #10583: URL: https://github.com/apache/seatunnel/pull/10583#issuecomment-5242243010
@danielnadean Thanks for digging into this — I independently re-checked the run and can confirm your diagnosis. Pulled the job logs for `kafka-connector-it (8, ubuntu-latest)` on this exact head (run `28919434367` on the fork), and the failure annotations are exactly `SeaTunnel job executed failed` / `Run SeaTunnel on spark failed`, with no relation to anything touched in this diff (this PR only adds Javadoc to `MultiTableWriterRunnable`, zero behavior change). Same pattern for `all-connectors-it-7` and `rocketmq-connector-it (11, ubuntu-latest)` — all three fail at the "run SeaTunnel on spark" e2e step, not at compile/unit-test. Your cross-PR observation checks out too, and I found a data point that strengthens it further: on #11642 (davidzollo's unrelated single-file API PR), the same three job names failed with the identical signature on the original run, but a same-day rerun (today) already came back green for `all-connectors-it-7` and `rocketmq-connector-it` with zero code changes — `kafka-connector-it` is still finishing there as I write this. That's about as clean a confirmation as we'll get that this is a transient Spark e2e infra flake and not something either PR's diff caused. One extra wrinkle worth flagging: this PR currently shows `mergeStateStatus: DIRTY` / `mergeable: CONFLICTING` against `dev` (the branch hasn't been synced since March/July and has drifted). So the lowest-cost path here probably isn't a bare job-level rerun of the three failing jobs — it's for @LiJie20190102 to rebase/merge latest `dev` into `doc_multi_table_writer_runnable` and push. That resolves the conflict blocking mergeability *and* triggers a brand-new CI run for free, which sidesteps the stale-run rerun-permission problem entirely. If a rerun is preferred instead (no rebase needed), any maintainer with fork access, or @LiJie20190102 directly, can use "Re-run failed jobs" from the fork's Actions run page. No action needed on the PR content itself — this is pure Javadoc and not implicated in the failure. -- 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]
