DanielLeens commented on PR #11569:
URL: https://github.com/apache/seatunnel/pull/11569#issuecomment-5548484714
Self-review note: this is my own PR, so GitHub blocks a formal review
submission on it — posting as a plain comment, consistent with every prior
round on this thread.
## CI follow-up on the current head (`b8d2dbad14ec`)
My last self-review said `Build` was `in_progress` on this head and that it
needed to finish before being called done from the CI side. It has now finished
red, so I traced the actual failure rather than leaving it unresolved:
- **`updated-modules-integration-test-part-2` failed on both JDK 8 and JDK
11.** Root cause, from the job log:
`org.apache.seatunnel.connectors.seatunnel.jdbc.internal.xa.XaGroupOpsImplIT` —
the exact test class this PR adds — errored on both its cases
(`testCommitFailurePropagatesThroughAggregatedCommitter`,
`testRecoverAndRollback`) before any XA logic ran, with:
```
com.github.dockerjava.api.exception.NotFoundException: Status 404:
{"message":"No such image: mysql:8.0.43"}
```
This is Testcontainers failing to create the `mysql:8.0.43` container — a
Docker image pull/availability issue on the runner, not an assertion failure in
the test itself. I checked whether this PR introduced that image tag: it didn't
— `mysql:8.0.43` is the standing `MYSQL_DOCKER_IMAGE` constant already used
across many pre-existing IT classes in `dev` (`JdbcMysqlIT`,
`SinkErrorToMysqlIT`, `TransformErrorToMysqlIT`, etc.), so this isn't a config
mistake specific to `XaGroupOpsImplIT`. Right after this failure, the very next
test class in the same job (`JdbcMariaDBIT`) successfully pulled and started
its own container a few seconds later, which points to a transient
registry/pull hiccup at that moment rather than a systemic image-availability
problem.
**Conclusion: not a defect in this PR's XA commit-failure-propagation
logic** — neither new test case got far enough to exercise
`XaGroupOpsImpl`/`wrapException` at all; they failed at container setup. This
carries forward my source-level conclusion unchanged: no blocker. A job-level
rerun of `updated-modules-integration-test-part-2` (both JDK legs) should clear
it without any code change.
**What's actually left before merge is still process, not code:** a
maintainer with write access needs to give a live, non-self approval, same as
every prior round.
--
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]
