Release v24.3.18 · redpanda-data/redpanda

Bug Fixes

Fixed an issue with consumer groups with manually assigned consumers. When an OffsetDeleteRequest was sent on such a group, a GROUP_SUBSCRIBED_TO_TOPIC error was returned. by @IoannisRP in #26705
Increase the default self check timeout from 5s to 10s to leave time to retry DNS lookups if they time out during a self check operation. by @pgellert in #26775
Resolves a memory leak scenario in node_status_backend by resetting connections which make no progress by @joe-redpanda in #26802
Return the correct error response if the RPC to the leader for deleting ACLs fails. by @BenPope in #26785
This fixes a bug in Redpanda's self-check functionality, where the self-check would occasionally fail with 'Uploaded key/payload could not be found in cloud storage item list.' despite the object being successfully uploaded. This issue occurred when testing against an Azure ABS tiered storage endpoint. by @pgellert in #26728
#26739 Fixes a bug in which a segment produced by adjacent merge compaction did not have its batch cache reset, leading to potentially stale reads in the storage layer. by @WillemKauf in #26741
#26820 Fixes a bug where data loss could occur during FPM w/ tiered storage disabled by @oleiman in #26821

Make segment download timeouts configurable in cloud cache hydration by @oleiman in #26779
PR #26744 [v24.3.x] [CORE-12729] debug/bundle: forward kubernetes env vars to rpk by @IoannisRP
PR #26815 [v24.3.x] [CORE-8805] dt/archival: Decrease manifest upload interval to avoid race by @oleiman
PR #26769 [v24.3.x] raft/heartbeat_manager: timeout memory leak fix by @joe-redpanda

Full Changelog: v24.3.17...v24.3.18