taoran92 opened a new pull request, #4400:
URL: https://github.com/apache/flink-cdc/pull/4400

   # What is the purpose of the change
   
     This PR fixes inconsistent source identity propagation in the distributed 
schema evolution flow.
   
     In distributed mode, `SchemaOperator` forwards schema change requests to 
the coordinator with the original upstream source partition via 
`SchemaChangeRequest(sourcePartition, sinkSubTaskId, schemaChangeEvent)`.
     However, when emitting the downstream `FlushEvent`, it previously used the 
current `SchemaOperator` subtask id instead of the original source partition.
   
     As a result, the same schema change could be identified by different ids 
in different parts of the pipeline:
   
     - `SchemaChangeRequest` / coordinator deduplication use the original 
source partition
     - `FlushEvent` / downstream flush handling use the `SchemaOperator` 
subtask id
   
     This makes the flush lineage inconsistent with the schema-change lineage 
in distributed topology. In particular, duplicated broadcast branches of the 
same upstream schema change cannot be consistently aligned
     by the original source partition.
   
   # Brief change log
   
     - Use `schemaChangeRequest.getSourceSubTaskId()` instead of 
`SchemaOperator`'s `subTaskId` when constructing distributed `FlushEvent`
     - Add a regression test in `SchemaEvolveTest` to verify that `FlushEvent` 
keeps the original source partition even when the `SchemaOperator` subtask id 
is different
     - Extend `DistributedEventOperatorTestHarness` so tests can inject a 
custom operator subtask index and reliably reproduce this scenario
   
   # Verifying this change
   
     This change is verified by unit tests:
   
     - 
`SchemaEvolveTest#testFlushEventUsesSourcePartitionInsteadOfSchemaOperatorSubtask`
   
     Does this pull request potentially affect one of the following parts:
   
     - Dependencies (does it add or upgrade a dependency): no
     - The public API, i.e. is any changed class annotated with 
`@Public`(`@PublicEvolving`): no
     - The serializers: no
     - The runtime per-record code paths (performance sensitive): no
     - Anything that affects deployment or recovery: no
   
   # Documentation
   
   Does this pull request introduce a new feature? no
   
   If yes, how is the feature documented? not applicable


-- 
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