Bug Fixes
- Addresses startup behavior in SR that may result in inconsistent SR state between nodes by @michael-redpanda in #25406
- Fixes a bug in which urgent garbage collection of multiple partitions may only consider one partition in the set. by @WillemKauf in #25275
- 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 #25665
- #25284 Fix a bug in group management behaviour when the last member of a group expires while there are pending
members in the group. by @pgellert in #25302 - #25523 kafka/server/group_manager: Prevent metrics::double_registration during group management. by @BenPope in #25525
- fixes a very rare situation in which Raft leader can enter into infinite loop trying to recover follower. by @mmaslankaprv in #25377
- PR #25417 [v24.2.x] pandaproxy/sr: Fix normalization not being applied on stored schemas by @IoannisRP
- PR #25445 [v24.2.x] c/rm_stm: correct error codes by @bashtanov
- PR #25466 [v24.2.x] pandaproxy/sr: Relax schema registry startup validity requirements by @IoannisRP
- PR #25477 [v24.2.x] c/backend: fix copy pasta error in learner offset calculation by @bharathv
- PR #25528 [v24.2.x] pandaproxy/sr: Re-purpose schema_registry_normalize_on_startup config by @IoannisRP
- PR #25543 [v24.2.x] s3: Add timeout to credentials fetching operation by @Lazin
- PR #25551 [v24.2.x] cluster/partition: Reword description for kafka_request_bytes_total. by @BenPope
- PR #25576 [v24.2.x] Conservative validation in by @mmaslankaprv
- PR #25584 [v24.2.x] config: relax schema_registry_always_normalize to not need restart by @IoannisRP
Improvements
- admin: Disallow a superuser from removing themself. by @BenPope in #25547
- PR #25154 [v24.2.x] kafka/server: fix case of use after move by @IoannisRP
Full Changelog: v24.2.20...v24.2.21