DanielLeens commented on issue #12599: URL: https://github.com/apache/seatunnel/issues/12599#issuecomment-5951993787
Classification: A / MySQL CDC schema nullability and data correctness. I checked current `dev` and #12597. #12597 changes JDBC metadata checks used by `AbstractJdbcSourceChunkSplitter`, so it prevents a nullable unique key from being selected for a parallel snapshot split. It does not change `MySqlSchema#parseSnapshotDdl`, the Debezium table metadata it returns, or the binlog value-conversion path. It can fix the snapshot-row omission in #12598, but it cannot by itself fix this issue's `NULL -> 0` binlog conversion. Please do not close #12599 as covered by #12597. The focused fix should restore database nullability for the Debezium-derived schema when a table has no real primary key, without changing a genuinely non-null primary-key column. Please prove it with one MySQL CDC regression that covers both snapshot and binlog records, a nullable single-column unique key, a composite unique key containing a nullable column, and controls for a real primary key and a non-null unique key. The sink assertion must distinguish SQL NULL from the type default `0`. -- 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]
