DanielLeens commented on issue #11034: URL: https://github.com/apache/seatunnel/issues/11034#issuecomment-4671702532
Thanks for the detailed reproduction. After checking the current Oracle CDC schema-loading path, this does look like a real connector-side bug rather than an Oracle 19c usage issue. The failing path is `OracleDialect.queryTableSchema -> OracleSchema.readTableSchema`. In the current implementation, the connector re-reads schema metadata and only keeps tables whose returned `TableId` exactly matches the cached `tableMap`. In a multi-schema CDB/PDB job like yours, if the returned identifiers do not line up exactly with the cached keys, the connector can end up throwing `Can't obtain schema for table ...` even though the table itself is valid and accessible. That direction is also consistent with your reproduction: single-table works, two tables in one schema work, but the combined multi-table / multi-schema path fails on the later schema load. I do not see a clean config-only workaround from the current code path, so this is worth keeping open as a real bug. If someone picks it up, the first place to instrument is the `OracleSchema.readTableSchema()` cache/matching path and the exact `TableId` values returned by `readSchema(...)` for `ORCLPDB.LIB_B.T_B1`. We've labeled this as `help wanted` as well, so contributions are very welcome. -- 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]
