wombatu-kun commented on PR #19200:
URL: https://github.com/apache/hudi/pull/19200#issuecomment-4894389403

   The `test-flink-1 (flink2.1)` failure here is an unrelated flake, not caused 
by this PR. The job died on 
`ITTestHoodieDataSource.testStreamReadFromSpecifiedCommitWithChangelog` with a 
`NullPointerException` in the Flink CDC read path 
(`CdcIterators.BaseImageIterator#hasNext`), a subsystem this PR does not touch. 
It reproduced on only that one shard while `test-flink-2 (flink2.1)` and every 
other Flink shard passed, which is the signature of a teardown timing race 
rather than a regression.
   
   Root cause: `CdcFileSplitsIterator` is drained on the Flink task thread but 
can be `close()`d from the split-fetcher thread during forced job teardown, 
racing on its iterator fields. This path only started being exercised by 
default after native CDC read (#19114) and the native-log-format default switch 
(#19118).
   
   Fix filed separately in #19202 (production thread-safety on the CDC read 
iterator plus a test-side teardown-race tolerance). Re-running that shard 
should pass in the meantime.
   


-- 
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]

Reply via email to