davsclaus commented on PR #25639:
URL: https://github.com/apache/camel/pull/25639#issuecomment-5407182680
Closing this in favor of #25645, which fixes the actual root cause.
The timeout increase here can't work: the failure is deterministic, not a
timing flake. `testReadinessWhenDown` calls `service.shutdown()` to simulate a
downed broker, but `KafkaHealthCheckTestSupport` uses
`KafkaServiceFactory.createSingletonService()`, and
`SingletonService.shutdown()` is a deliberate no-op ("Ignoring shutdown request
... will be shutdown via JVM shutdown hook"). So the broker never actually goes
down — the consumer stays connected, the readiness check keeps reporting UP,
and the `await(...DOWN...)` times out at any window (confirmed at both 20s and
45s in CI).
This was introduced by CAMEL-24387 (#22294), which migrated these
health-check ITs to the singleton service. #25645 correctly reverts just this
base class back to `createService()`, whose `shutdown()` genuinely stops the
container.
Verified locally with a real Kafka container:
- singleton (this PR's state): FAIL — `ConditionTimeout ... within 20
seconds`
- `createService()` (#25645): PASS in ~4.8s
Also filing a follow-up: under `group.protocol=consumer`
(AsyncKafkaConsumer), the reflection in the readiness check can't reach
`ConsumerNetworkClient`, so it would never report DOWN — a separate latent
issue.
_Claude Code on behalf of davsclaus_
--
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]
