The GitHub Actions job "Flink CDC CI" on flink-cdc.git/master has failed.
Run started by GitHub user lvyanquan (triggered by lvyanquan).

Head commit for run:
101143217237f9e4021de43ecb15863568f973d4 / YasuoStudyJava 
<[email protected]>
[FLINK-41107][pipeline-connector] Honor accepted schema evolution types in Hudi 
and MaxCompute sinks (#4599)

HudiMetadataApplier and MaxComputeMetadataApplier relied on the default 
implementations of MetadataApplier#setAcceptedSchemaEvolutionTypes and 
MetadataApplier#acceptsSchemaEvolutionType, which discard the configured event 
types and accept every schema change event. As a result, 
sink.exclude.schema.changes / sink.include.schema.changes (and therefore 
schema.change.behavior: lenient) had no effect for these two sinks: excluded 
truncate.table and drop.table events were still forwarded to the sink, which 
could truncate or drop target tables on MaxCompute and fail Hudi jobs.

This closes the gap:
- HudiMetadataApplier keeps the enabled schema evolution types and declares 
only the event types it can actually apply: create.table, add.column, 
alter.column.type, drop.column and rename.column.
- MaxComputeMetadataApplier keeps the enabled schema evolution types so that 
events excluded by the pipeline configuration are no longer applied.
- Add unit tests covering the default supported types and the filtering 
performed after setAcceptedSchemaEvolutionTypes is called.

Co-authored-by: chenxin49 <[email protected]>

Report URL: https://github.com/apache/flink-cdc/actions/runs/41610200452

With regards,
GitHub Actions via GitBox

Reply via email to