BIOINSu commented on PR #24128:
URL: https://github.com/apache/flink/pull/24128#issuecomment-1905451090

   @libenchao Thanks for the review comments. In fact, the "Deal with changelog 
messages" is necessary. 
   
   In the previous version of the code, since consuming the changelog stream 
will add a `ChangelogNormalize` node after the Source node, the shuffle method 
between the two nodes is Hash Shuffle. Therefore, i hope to rely on such an 
association relationship to handle changelog data.
   
   However, since setting source parallelism and processing changelog data 
should be two independent features, in the new version of the code, I added 
`PartitionTransformation` after the `SourceTransformationWrapper` to ensure 
that the source and downstream nodes are Hash Shuffle after adjusting the 
parallelism. Since `PartitionTransformation` does not essentially generate 
actual physical nodes and selects the first `PartitionTransformation` 
corresponding partitioner as the real partitioner when there are multiple 
consecutive `PartitionTransformation`, this implementation will not affect the 
existing logic.


-- 
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