Bug Fixes
- Apply the
controller_mutation_ratequota to all parts of a partition mutation request using a consistent timestamp to prevent scheduling delays during request processing from affecting the throttling outcome. by @pgellert in #25632 - Fix an assert crash that happens after a node rejoins the cluster with the same node ID after its disk was erased. by @ztlpn in #25752
- This fixes a bug where kafka responses could get out-of-order with respect to request order on the same connection. This would surface as CorrelationIdMismatchException exceptions at the kafka client side. by @pgellert in #25664
Improvements
- Improve produce error response when produce requests are rejected due to full disk. by @pgellert in #25720
- #25621 Kafka/server: Improve logging for group removal due to retention. by @BenPope in #25623
Full Changelog: v24.3.9...v24.3.10