DanielLeens commented on issue #12565:
URL: https://github.com/apache/seatunnel/issues/12565#issuecomment-5917739950

   Classification: A / CDC base restore filtering and incremental delivery.
   
   I checked current `dev` (`53733a112c9e935c5d97f7f24be249caa43b4fd5`). 
`IncrementalSourceStreamFetcher.configureFilter()` rebuilds per-table 
high-watermark state from `completedSnapshotSplitInfos`, while an empty set 
applies one startup offset to every captured table. That makes the serialized 
incremental-split contents the decisive evidence for this report; the 
configuration alone cannot prove which old-table records were filtered after 
the table list changed.
   
   Before a production fix or `help wanted` decision, please provide a minimal 
MySQL CDC regression case that captures all of the following: the exact commit, 
the saved `IncrementalSplit` (table IDs, startup offset, completed-snapshot 
split high watermarks), the table-list change, and ordered source/sink records 
for both an old and the newly added table across the restore window. The 
assertion must demonstrate an old-table change after the saved offset is lost 
while the new-table snapshot succeeds. This will distinguish split-state 
reconstruction from a connector-specific offset or sink problem and give a 
testable acceptance boundary.
   


-- 
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]

Reply via email to