alanlau28 opened a new pull request, #23891: URL: https://github.com/apache/kafka/pull/23891
Jira: https://issues.apache.org/jira/browse/KAFKA-19473 A broker answers a fetch request at version 0 to 2 with an empty record set when the first batch at the fetch offset is larger than the per-partition limit. Older consumers rely on getting a partial batch in this case to raise RecordTooLargeException, so instead they keep fetching the same offset forever. `LazyDownConversionRecords` only reserved room for the overflow batch when the records held a complete batch, which a slice holding only part of a batch does not, so it now checks whether the records hold any bytes. Down-conversion was removed in 4.0, so this change targets 3.9. -- 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]
