SEZ9 commented on PR #12298: URL: https://github.com/apache/seatunnel/pull/12298#issuecomment-5723587730
A correction to my own CI analysis on this PR, and a short status note. Earlier I wrote that `SplitClusterFaultToleranceIT.testStreamJobCancelResolvesWhenWorkerCrashesBeforeCancelAck` fails 8/8 on #12299 while passing 3/3 on #12298, and floated suite size (204 vs 203 tests) as the only correlate. **That was a measurement error on my side.** I read only the newest workflow run on the #12298 branch, and GitHub carries unchanged jobs forward into each `rerun --failed` attempt with an identical `started_at`, so one pass looked like eight. Counting the earlier runs on that branch, #12298 fails it **4 of 8** real executions — all at 204 tests, which falsifies the suite-size idea outright. Corrected totals for JDK 8 / Linux: #12298 4/8, #12299 8/8, and 1/1 on a branch that is `dev` @ `75b60fa14` plus a comment-only change. **13 failures in 17 real executions across three code bases**, i.e. a base-level race with a high hit rate and no branch correlation. Details and the per-attempt table are in #12353, where @DanielLeens has since identified the mechanism (`failedTaskNum > 0` taking priority in `SubPlan.getPipelineEndState()`, reachable when a worker's `FAILED` callback lands inside the cancel-ack window) and #12311 looks like the fix. Nothing here changes this PR's own status: everything it touches is green, and the remaining red is pre-existing on `dev`. For the record, the current failing jobs are - `all-connectors-it-2` (8 and 11) — `OpengaussCDCIT.testAddFieldWithRestore`, #12344, 40 failures in 40 real executions, never observed passing - `engine-v2-it (8)` — the race above (#12353), or `BackpressureSlowSinkIT` on more recent attempts (#12313 is the fix for that one) I've stopped rerunning. A survey I posted on #12344 shows 15 of the 40 most recently updated open PRs sitting at `Build=FAILURE`, with every Zeta/engine PR in that window red and every connector-only PR green — including #12311 and #12313, the fixes for the other two tests. Rerunning cannot clear any of it, so this PR stays blocked on #12344 rather than on anything in its diff. -- 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]
