DanielLeens commented on PR #12163: URL: https://github.com/apache/seatunnel/pull/12163#issuecomment-5672693077
Thanks for the trace, @zhangshenghang — that's solid work matching the signatures against sibling PR runs. I checked the rerun (`34791705815`) after it finished, though, and there's more in it than the two flakes you diagnosed, so wanted to share the full picture before we call this clean: - **`OpengaussCDCIT.testAddFieldWithRestore`** and **`SplitClusterFaultToleranceIT.testStreamJobCancelResolvesWhenWorkerCrashesBeforeCancelAck`**: confirmed, still failing with the identical signatures you found (`all-connectors-it-2` on both JDK 8 and 11, `engine-v2-it (8)`). Neither test touches `connector-cdc-base` or the Zeta enumerator-context change this PR makes, so I agree these are unrelated. `all-connectors-it-1` passed on the rerun, consistent with the flake theory. - **New in this rerun — `all-connectors-it-7` (both JDK 8 and 11)**: `S3FileConnectDryRunIT` now fails with `pull access denied for minio/minio` (`Can't get Docker image: RemoteDockerImage(imageName=minio/minio:RELEASE.2024-06-13T22-53-53Z...)`). This is the exact Docker Hub image-removal issue, and it's already fixed upstream: #12302, "[Fix][CI] Use MinIO mirror for S3 dry-run tests," merged into `dev` a couple hours ago (`2026-09-14T23:04:39Z`), touching exactly `S3FileConnectDryRunIT.java`. Your branch is currently 5 commits behind `dev` (compare status: `diverged`), so it predates that fix. No overlap with this PR's diff either way. - **New in this rerun — `engine-v2-it (8, ubuntu-latest)`**: also failed on `BackpressureSlowSinkIT.testCheckpointsKeepCompletingUnderSustainedBackpressure` — `expected at least 3 additional checkpoints to complete during the 90s sustained backpressure window, only observed 0`. This test exercises generic sink backpressure/checkpoint scheduling with zero call-chain overlap with the CDC snapshot-split ack path or `SeaTunnelSplitEnumeratorContext` that this PR touches, and it ran back-to-back with two other long-running tests in the same job (right after a 529s test and then a separate 119.9s test), which points more at CI-runner contention than a functional regression. That said, I haven't independently corroborated this specific signature against other concurrent runs the way you did for the CDC test, so I'd call it unconfirmed rather than a settled flake for now. Given one of the four failures has a confirmed upstream fix already sitting in `dev` and the branch is diverged from `dev` by 5 commits, the lowest-cost next step is: sync the latest `dev` (which picks up #12302) and rerun CI. If `BackpressureSlowSinkIT` or the other two show up again after that, worth a closer look then — but I wouldn't expect any of the four to trace back to this PR's diff. My prior approval on this head's source-side content still stands — this is purely a CI-status update, no new code-side blocker from my side. -- 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]
