SEZ9 commented on PR #12311: URL: https://github.com/apache/seatunnel/pull/12311#issuecomment-5861513756
@DanielLeens @davidzollo CI follow-up on September 28: our current #12299 and #12454 Java 8 engine jobs both hit `SplitClusterFaultToleranceIT.testStreamJobCancelResolvesWhenWorkerCrashesBeforeCancelAck` (`CANCELED` expected, `FAILED` observed). Their current heads keep this correction separate, and their failed-only retry monitors are active. I re-read this PR at `db0516982cad`. The current diff still uses the vertex-only `resolveLostMemberState(ExecutionState)` helper. The job/pipeline-status gating and synchronized read/decision discussed in your September 26 replies are not present yet. Could you prioritize that revision, including the master-failover and not-yet-cancelled sibling cases, and post the revised head for review? The acknowledged edge cases need to remain part of the validation; the older patch should not be treated as covering them. There is also a concrete CI queue issue: Build `36327192530` on the same head still has 6 successful jobs, 73 queued jobs, 14 conditional skips, and no running or failed jobs. This is unchanged from my September 27 18:22 UTC observation through September 28 01:00 UTC. Could you investigate/unblock that fork Build queue while preparing the revision? The public run/job data does not identify why it is stuck. For additional evidence, I reproduced the worker-cancel assertion on unmodified `dev@146a1b5c510e` locally with Java 8/two processors. An isolated combination of common head `3080371ce733` with the original `ca60502b9e21` patch passed Spotless, full-reactor verification, and 53 selected unit/integration cases, including the worker-cancel test and other failover paths. That is evidence for the scenarios exercised, not validation of the acknowledged missing edge cases or a reason to merge the older mapping as-is. Neither of our open PRs includes that patch. -- 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]
