nielifeng opened a new pull request, #11503: URL: https://github.com/apache/seatunnel/pull/11503
### Purpose Fixes CDC schema recovery after a Zeta task failover when DDL changes were captured in a completed checkpoint. Closes #10461. ### Root cause - IncrementalSourceReader restored checkpoint tables only when the deprecated checkpointDataType field was non-null. - Debezium historyTableChanges were not restored into the deserializer. - Even after restoring the deserializer, the source collector, transforms, and sink writers could keep the original runtime row type. ### Changes - Restore non-empty checkpointTables and historyTableChanges independently of checkpointDataType. - Replace stale Debezium table history with checkpoint history. - Emit RestoreTableSchemaEvent before the first restored data record. - Propagate the restored runtime schema through transforms and schema-aware sinks. - Rebuild sink runtime serializers and writers without applying physical DDL a second time. - Preserve compatibility with legacy checkpoints that contain empty restore state. ### Validation - RestoreTableSchemaEventTest: passed. - CDC restore unit tests: 6 passed, 0 failed. - Production compilation passed for API, CDC Base, JDBC, Databend, Doris, Elasticsearch, File, Iceberg, Paimon, StarRocks, and Transforms V2. - Transforms V2 and File Base module test suites passed during cross-module validation. - Spotless checks passed for all changed modules. - Real MySQL CDC to Console validation: two consecutive Worker failures and recoveries, restored rows retained the new column, 139 checkpoints completed, 0 failed, 2 restored. - Real MySQL CDC to JDBC validation: physical ADD COLUMN executed once, runtime schema restored without duplicate DDL, 66 checkpoints completed, 0 failed, 1 restored. -- 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]
