SEZ9 commented on issue #12482: URL: https://github.com/apache/seatunnel/issues/12482#issuecomment-5842972596
Thanks for the detailed analysis and data. Splitting the second host-JDK leg out of PR connector E2E runs while keeping unit tests, engine E2E and transform E2E on both JDKs makes sense, since the engines under test run on their own container JVMs. Two things would help before this goes to dev@: - The 66-run sample is from a single fork. Extending it to the larger sample you mention would give more confidence that none of the JDK-11-only connector failures are genuine host-JDK incompatibilities (e.g. Testcontainers client or driver classloading differences). The `RocketMqIT` assertion and `JdbcPhoenixIT` class setup failures are worth a closer look for that reason. - The nightly is the backstop for the dropped leg, so I'd treat the #12467 concurrency fix as a prerequisite for switching PR runs to one JDK. On the host JDK for the single PR leg, using the lower supported JDK (8 now, 11 after #11545) seems reasonable. The optional `full-ci` label for restoring both JDKs on a PR is also a good addition. Let's continue on dev@. <!-- streview-comment:1328 --> -- 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]
