Merge pull request #9495 from rabbitmq/mk-3.12.5-release-notes

3.12.5 release notes

Author: michaelklishin

release-notes/3.12.5.md

@@ -47,6 +47,67 @@ Release notes can be found on GitHub at [rabbitmq-server/release-notes](https://
 
    GitHub issue: [#9222](https://github.com/rabbitmq/rabbitmq-server/pull/9222)
 
+ * Avoids a potential exception in the credit flow subsystem.
+
+   Contributed by @Ayanda-D.
+
+   GitHub issue: [#9433](https://github.com/rabbitmq/rabbitmq-server/pull/9433)
+
+ * Classic queues v2 (CQv2) that had priorities enabled were reporting their storage version
+   incorrectly.
+
+   Contributed by @gomoripeti (CloudAMQP).
+
+   GitHub issue: [#9370](https://github.com/rabbitmq/rabbitmq-server/issues/9370)
+
+#### Enhancements
+
+ * Channels that had many thousands of consumers (usually due to a consumer leak in the application)
+   now consume a lot less CPU resources when the channel is closed.
+
+   As part of this optimization, individual `consumer.deleted` internal events are no
+   longer emitted when a channel is closed, only a `channel.closed` event is. This also
+   help reduce audit event log size, since those `consumer.deleted` events provided
+   no useful information in this context.
+
+   Contributed by @SimonUnge (AWS).
+
+   GitHub issue: [#9356](https://github.com/rabbitmq/rabbitmq-server/pull/9356)
+
+ * Initial forward compatibility with later Erlang 26.x and 27.0 releases.
+
+   GitHub issue: [#9485](https://github.com/rabbitmq/rabbitmq-server/pull/9485)
+
+ * Nodes now log boot time at info level instead of debug. This piece of information
+   can be useful during root cause analysis.
+
+   Contributed by @johanrhodin (CloudAMQP).
+
+   GitHub issue: [#9466](https://github.com/rabbitmq/rabbitmq-server/pull/9466)
+
+ * Channel interceptors (defined by plugins) now can return one more type of errors.
+
+   Contributed by @Ayanda-D.
+
+   GitHub issue: [#9459](https://github.com/rabbitmq/rabbitmq-server/pull/9459)
+
+
+### CLI Tools
+
+#### Bug Fixes
+
+ * `rabbitmqctl delete_queue` now handles more queue replica failure scenarios
+   (that were previously handled by client operations and the HTTP API but not CLI tools).
+
+   Contributed by @Ayanda-D.
+
+   GitHub issue: [#9324](https://github.com/rabbitmq/rabbitmq-server/pull/9324)
+
+ * `rabbitmq-streams delete_replica` now handles scenarios where the replica is hosted
+   on a node that is not accessiable (reachable).
+
+   GitHub issue: [#9282](https://github.com/rabbitmq/rabbitmq-server/issues/9282)
+
 
 ### MQTT Plugin
 
@@ -58,9 +119,31 @@ Release notes can be found on GitHub at [rabbitmq-server/release-notes](https://
    GitHub issue: [#9281](https://github.com/rabbitmq/rabbitmq-server/pull/9281)
 
 
+### AMQP 1.0 Plugin
+
+#### Bug Fixes
+
+ * Listing AMQP 1.0 connections in a system that had both AMQP 1.0 and AMQP 0-9-1 clients connected
+   resulted in exceptions logged and partial results returned to CLI tools.
+
+   GitHub issue: [#9371](https://github.com/rabbitmq/rabbitmq-server/issues/9371)
+
+
+### Prometheus Plugin
+
+#### Enhancements
+
+ * The plugin now exposes a new gauge, `rabbitmq_unreachable_cluster_peers_count`, that indicates how many cluster
+   peers **cannot** be reached by this node.
+
+   Contributed by @gomoripeti (CloudAMQP).
+
+   GitHub issues: [#2508](https://github.com/rabbitmq/rabbitmq-server/issues/2508), [#9465](https://github.com/rabbitmq/rabbitmq-server/pull/9465)
+
+
 ## Dependency Upgrades
 
- * `osiris` was upgraded to [`1.6.4`](https://github.com/rabbitmq/osiris/tags)
+ * `osiris` was upgraded to [`1.6.7`](https://github.com/rabbitmq/osiris/tags)
 
 
 ## Dependency Upgrades