JNSimba opened a new pull request, #708:
URL: https://github.com/apache/doris-flink-connector/pull/708

   ## Problem Summary
   
   A new incremental source job currently ignores offsets previously published 
to `source.binlog.offset-table`: `initial` repeats its snapshot and 
`from-timestamp` reuses the configured start time.
   
   Read the configured consumer's saved boundary once during new Source 
initialization and start directly in the incremental phase when a valid offset 
exists. Restored Flink Source state remains authoritative and does not query 
the external table. Without a saved record, `initial` performs its first 
snapshot, `from-timestamp` uses its configured time, and `latest` with an 
offset table fails rather than silently skipping history. Connection, 
permission, duplicate-record, and invalid-timestamp failures are propagated.
   
   This change reuses the existing offset configuration and publication path. 
It does not change tablet scheduling, Reader/Enumerator coordination, or 
checkpoint formats.
   
   ## Checklist
   
   1. Does it affect the original behavior: Yes, new jobs with a configured 
offset table prefer a saved offset, including `from-timestamp`.
   2. Has unit tests been added: Yes, startup precedence, empty records, 
invalid records, and query/connection failures; integration cases cover restart 
without Flink state and `initial` without a repeated snapshot.
   3. Has document been added or modified: Yes, external-offset startup rules 
and SELECT permission requirements.
   4. Does it need to update dependencies: No.
   5. Are there any changes that cannot be rolled back: No.
   
   ## Validation
   
   - Static review and Java formatting completed.
   - `git diff --check` passed.
   - Compilation, unit tests, and integration tests have not been run.
   
   External offsets restore only Source progress, not Flink operator state or 
Sink transactions. Restart can replay records; this is not an end-to-end 
exactly-once recovery mechanism.
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to