davidzollo commented on PR #11198:
URL: https://github.com/apache/seatunnel/pull/11198#issuecomment-5242315156
Thanks @davidzollo for the approval, and thanks for sticking with this one
@srijan-singh.
Re-checked the live state on the current head (`391b54ffe89d`) just now:
`Build` is still **FAILURE**, so this isn't merge-ready yet — CI hasn't
actually turned green.
Tracing the outer Apache "Build" pointer to the real fork run, two jobs are
failing:
- `all-connectors-it-2 (8, ubuntu-latest)` — job step "run connector-v2
integration test (part-2)"; the check-run annotations point to a Spark-engine
job execution failure ("SeaTunnel job executed failed" / "Run SeaTunnel on
spark failed") inside that aggregated multi-connector partition, not anything
Couchbase-specific.
- `kafka-connector-it (11, ubuntu-latest)` — job step "run kafka connector
integration test", exit code 1.
On `kafka-connector-it`: this PR has hit this exact lane before (see the
2026-07-04 comment thread), and the failing test that round was
`KafkaIT.testKafkaToKafkaExactlyOnceOnStreaming` with a
`ConditionTimeoutException` — a known-flaky Kafka IT, unrelated to the
Couchbase sink code. Given the pattern, my working assumption is this is the
same pre-existing flake rather than a regression from the current head, but I
haven't pulled today's raw job log far enough to confirm the specific failing
test name on this run, so I'd treat that as likely rather than confirmed.
One thing worth a deliberate double-check rather than assuming pure flake:
this PR also modifies the shared `seatunnel-e2e-common` test container
(`SeaTunnelContainer.java`'s thread-leak exemption logic, plus the new
`SeaTunnelContainerThreadExemptionTest`), which is a transitive test dependency
of essentially every connector-v2 E2E/IT module — including
`kafka-connector-it`. I already reviewed that change on its own terms in my
last pass and didn't flag it, but given it's a shared file and the failure is
in exactly that class of module, it's worth a quick sanity check that the
thread-check assertion isn't the actual cause on this run before writing it off
as unrelated flakiness.
To be clear on my own standing here: my last full review found no remaining
source-side blockers (only the three non-blocking Low/Medium items already on
record), so from my side this is purely a CI-gate question now, not a code
question. I only have comment-level access, so I can't formally move this to
approved myself — I'd support merging once `Build` is actually green on this
head and davidzollo's approval still applies to that green run.
--
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]
