Improved

JetStream

• The scan for the last sourced message sequence when setting up a subject-filtered source is now considerably faster (#7553)
• Consumer interest checks on interest-based streams are now significantly faster when there are large gaps in interest (#7656)
• Creating consumer file stores no longer contends on the stream lock, improving consumer create performance on heavily loaded streams (#7700)
• Recalculating num pending with updated filter subjects no longer gathers and sorts the subject filter list twice (#7772)
• Switching to interest-based retention will now remove no-interest messages from the head of the stream (#7766)

MQTT

• Retained messages will now work correctly even when sourced from a different account and has a subject transform (#7636)

Fixed

General

• WebSocket connections will now correctly limit the buffer size during decompression (#7625, thanks to Pavel Kokout at Aisle Research)
• The config parser now correctly detects and errors on self-referencing environment variables (#7737)
• Internal functions for handling headers should no longer corrupt message bodies if appended (#7752)

JetStream

• A protocol error caused by an invalid transform of acknowledgement reply subjects when originating from a gateway connection has been fixed (#7579)
• The meta layer will now only respond to peer remove requests after quorum has been reached (#7581)
• Invalid subject filters containing non-terminating full wildcard no longer produce unexpected matches (#7585)
• A data race when creating a stream in clustered mode has been fixed (#7586)
• A panic when processing snapshots with missing nodes or assignments has been fixed (#7588)
• When purging whole message blocks, the subject tracking and scheduled messages are now updated correctly (#7593)
• The filestore will no longer unexpectedly lose writes when AsyncFlush is enabled after a process pause (#7594)
• The filestore now will process message removal on disk before updating accounting, which improves error handling (#7595, #7601)
• Raft will no longer allow peer-removing the one remaining peer (#7610)
• A data race has been fixed in the stream health check (#7619)
• Tombstones are now correctly written for recovering the sequences after compacting or purging an almost-empty stream to seq 2 (#7627)
• Combining skip sequences and compactions will no longer overwrite the block at the wrong offset, correcting a corrupt record state error (#7627)
• Compactions that reclaim over half of the available space now use an atomic write to avoid losing messages if killed (#7627)
• Filestore compaction should no longer result in no idx present cache errors (#7634)
• Filestore compaction now correctly adjusts the high and low sequences for a message block, as well as cleaning up the deletion map accordingly (#7634)
• Potential stream desyncs that could happen during stream snapshotting have been fixed (#7655)
• Raft will no longer allow multiple membership changes to take place concurrently (#7565, #7609)
• Raft will no longer count responses from peer-removed nodes towards quorum (#7589)
• Raft quorum counting has been refactored so the implicit leader ack is now only counted if still a part of the membership (#7600)
• Raft now writes the peer state immediately when handling a peer-remove to ensure the removed peers cannot unexpectedly reappear after a restart (#7602)
• Raft will no longer allow peer-removing the one remaining peer (#7610)
• Add peer operations to Raft can no longer result in disjoint majorities (#7632)
• Raft groups should no longer readmit a previously removed peer if a heartbeat occurs between the peer removal and the leadership transfer (#7649)
• Raft single node elections now transition into leader state correctly (#7642)
• R1 streams will no longer incorrectly drift last sequence when exceeding limits (#7658)
• Deleted streams are no longer wrongfully revived if stalled on an upper-layer catchup (#7668)
• A panic that could happen when receiving a shutdown signal while JetStream is still starting up has been fixed (#7683)
• JetStream usage stats now correctly reflect purged whole blocks when optimising large purges (#7685)
• Recovering JetStream encryption keys now happens independently of the stream index recovery, fixing some cases where the key could be reset unexpectedly if the index is rebuilt (#7678)
• Non-replicated file-based consumers now detect corrupted state on disk and are deleted automatically (#7691)
• Raft no longer allows a repeat vote for the same term after a stepdown or leadership transfer (#7698)
• Replicated consumers are no longer incorrectly deleted if they become leader just as JetStream is about to shut down (#7699)
• Fixed an issue where a single truncated block could prevent storing new messages in the filestore (#7704)
• Fixed a concurrent map iteration/write panic that could occur on WorkQueue streams during partitioning (#7708)
• Fixed a deadlock that could occur on shutdown when adding streams (#7710)
• A data race on mirror consumers has been fixed (#7716)
• JetStream no longer leaks subscriptions in a cluster when a stream import/export is set up that overlaps the $JS.> namespace (#7720)
• The filestore will no longer waste CPU time rebuilding subject state for WALs (#7721)
• Configuring cluster_traffic in config mode has been fixed (#7723)
• Subject intersection no longer misses certain subjects with specific patterns of overlapping filters, which could affect consumers, num pending calculations etc (#7728, #7741, #7744, #7745)
• Multi-filtered next message lookups in the filestore can now skip blocks when faster to do so (#7750)
• The binary search for start times now handles deleted messages correctly (#7751)
• Consumer updates will now only recalculate num pending when the filter subjects are changed (#7753)
• Consumers on replicated interest or workqueue streams should no longer lose interest or cause desyncs after having their filter subjects updated (#7773)
• Interest-based streams will no longer start more check interest state goroutines when there are existing running ones (#7769)

MQTT

• The maximum payload size is now correctly enforced for MQTT clients (#7555, thanks to @yixianOu)
• Fixed a panic that could occur when reloading config if the user did not have permission to access retained messages (#7596)
• Fixed account mapping for JetStream API requests when traversing non-JetStream-enabled servers (#7598)
• QoS0 messages are now mapped correctly across account imports/exports with subject mappings (#7605)
• Loading retained messages no longer fails after restarting due to last sequence checks (#7616)
• A bug which could corrupt retained messages in clustered deployments has been fixed (#7622)
• Permissions to $MQTT. subscriptions are now handled implicitly, with the exception of deny ACLs which still permit restriction (#7637)
• A bug where QoS2 messages could not be retrieved after a server restart has been fixed (#7643)

Complete Changes

https://github.com/nats-io/nats-server/compare/v2.11.11...v2.11.12