The GitHub Actions job "Flink CDC CI" on 
flink-cdc.git/fix/iceberg-duplication-same-checkpoint has failed.
Run started by GitHub user spoorthibasu (triggered by yuxiqian).

Head commit for run:
a6183525619eb2b48be4d251813dd4062498cb0e / Spoorthi Basu 
<[email protected]>
[iceberg] Fix batchIndex sync and parallel subtask grouping in Iceberg sink

Address parallelism issues identified during review:
- Writer: Advance tableBatchIndexMap before the writer == null guard so all 
subtasks stay in sync when a subtask has no data for the table at schema-change 
time
- Writer: Skip flushTableWriter on initial CreateTableEvent since no data has 
been written yet and there is nothing to split
- Committer: Group WriteResultWrappers by batchIndex using a TreeMap, so 
wrappers from different subtasks with the same batchIndex are merged into a 
single Iceberg snapshot instead of being committed separately

Tests added:
- testBatchIndexInSyncWhenSubtaskHasNoWriterAtSchemaChange
- testNoDuplicateWithParallelSubtasksMissingPreSchemaChangeData
- testSameBatchIndexFromTwoSubtasksMergedIntoOneSnapshot
- testNoDuplicateWithMixedDataAcrossSubtasksAndMultipleSchemaChanges

Report URL: https://github.com/apache/flink-cdc/actions/runs/24018734717

With regards,
GitHub Actions via GitBox

Reply via email to