DanielLeens commented on code in PR #12288:
URL: https://github.com/apache/seatunnel/pull/12288#discussion_r4254913477
##########
seatunnel-e2e/seatunnel-engine-e2e/connector-seatunnel-e2e-base/src/test/java/org/apache/seatunnel/engine/e2e/CheckpointCoordinatorFailoverIT.java:
##########
@@ -823,14 +845,27 @@ public void
testStreamJobFailsAfterCheckpointTriggerDispatchFailure() throws Exc
Assertions.assertEquals(JobStatus.FAILED, jobResult.getStatus());
Assertions.assertNotNull(
jobResult.getError(), "a FAILED job should carry a
non-null error message");
+ // Either reason is a correct detection of the same injected fault
-- see the
+ // "Which failure path actually fires is a genuine, harmless race"
section of the
+ // class javadoc above for why both
CheckpointManager#sendOperationToMemberNode call
+ // sites (checkpoint-trigger barrier dispatch and post-completion
notify) can observe
+ // the removed ownedSlotProfilesIMap entry depending on scheduling
timing.
Assertions.assertTrue(
jobResult
- .getError()
-
.contains(CheckpointCloseReason.CHECKPOINT_INSIDE_ERROR.message()),
+ .getError()
+ .contains(
+
CheckpointCloseReason.CHECKPOINT_INSIDE_ERROR.message())
+ || jobResult
Review Comment:
Reconciled against the current `dev` head: the deterministic replacement,
#12449, was merged as `146a1b5c510e1c33010a92981ae1afc8139c624b` on 2026-09-27.
`dev` now waits for `pendingCounter == 0` and removes the fault under the
coordinator lock, so it closes the race identified here without this branch’s
production-only `getLatestCompletedCheckpointId()` accessor. The loop evidence
in this thread also showed this branch remained flaky. I am therefore resolving
this thread and closing #12288 as superseded rather than resolving its merge
conflict.
--
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]