zhongyujiang opened a new pull request #4117: URL: https://github.com/apache/iceberg/pull/4117
This pr fix unit test TestFlinkSink#testHashDistributeMode which fails occassionally in Flink CI, have been discussed a lot in #2989 and #3365. I think the root cause is the way notifyCheckpointComplete works in IcebergFilesCommitter: https://github.com/apache/iceberg/blob/d43cb4c296f8096300b43fe830fe6c4fb085aaa5/flink/v1.14/flink/src/main/java/org/apache/iceberg/flink/sink/IcebergFilesCommitter.java#L182-L195 https://github.com/apache/iceberg/blob/ca46fc93ae9a349e3920d0519ccf0f41e7210d44/flink/v1.14/flink/src/main/java/org/apache/iceberg/flink/sink/IcebergFilesCommitter.java#L261-L277 As show above, results of multiple ckpts may be merged into one commit in streaming mode, and the checkpoint interval (400 ms) here is rather small, which makes this situation very likely. Increasing checkpoint interval would reduce such failure, but it cannot be completely eliminated in theory. So I simply made this unit test only apply for batch mode, which is enough to validate Hash distribute mode in my opinion. @openinx @szehon-ho could help review this? thanks! -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
