Zhuoxi2000 commented on PR #290:
URL: 
https://github.com/apache/flink-connector-kafka/pull/290#issuecomment-5443037466

   I looked at how #293 interacts with the lazy consumer, and I think the two 
changes compose cleanly:
   
   * #293 avoids touching the consumer during offset reconciliation. It uses 
`lastFetchedOffsets` / `lastKnownPositions`, which are populated from 
`fetch()`. Both fetching and offset commits run on the split fetcher executor, 
so this doesn’t introduce any new cross-thread access.
   * With lazy creation, a commit before the first fetch creates the consumer 
on that same fetcher thread, which is exactly what FLINK-36434 wants. Since 
nothing has been fetched yet, there are no known positions to reconcile, so the 
offsets pass through unchanged.
   * The overlap is mostly mechanical. If #293 lands first, the rebase here 
should just change `consumer.commitAsync(...)` to 
`ensureConsumer().commitAsync(...)`.
   
   Happy to rebase either way once the landing order is settled with @dalelane.
   


-- 
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