- When a watch is closed by an HTTP2 load balancer and we are told to go away, skip printing the message to stderr by default.
- Spedup kubectl by >10 when calling out to kube-apiserver for discovery information. ([#73345](https://github.com/kubernetes/kubernetes/pull/73345), [@sttts](https://github.com/sttts))

* Fix watch to not send the same set of events multiple times causing watcher to go back in time ([#73845](https://github.com/kubernetes/kubernetes/pull/73845), [@wojtek-t](https://github.com/wojtek-t))

Renamed FeatureGate RequestManagement to APIPriorityAndFairness....

* kube-apiserver: fixed a bug that could cause a goroutine leak if the apiserver encountered an encoding error serving a watch to a websocket watcher ([#84693](https://github.com/kubernetes/kubernetes/pull/84693), [@tedyu](https://github.com/tedyu))

* kube-apiserver: fixed a bug that could cause a goroutine leak if the apiserver encountered an encoding error serving a watch to a websocket watcher ([#84693](https://github.com/kubernetes/kubernetes/pull/84693), [@tedyu](https://github.com/tedyu))

* Fix concurrent map access panic ([#65328](https://github.com/kubernetes/kubernetes/pull/65328), [@dashpole](https://github.com/dashpole))
    * Don't watch .mount cgroups to reduce number of inotify watches
    * Fix NVML initialization race condition
    * Fix brtfs disk metrics when using a subdirectory of a subvolume

- The watch API endpoints prefixed with `/watch` are deprecated and will be removed in a future release. These standard method for watching resources (supported since v1.0) is to use the list API endpoints with a `?watch=true` parameter. All client-go clients have used the parameter method since v1.6.0. ([#65147](https://github.com/kubernetes/kubernetes/pull/65147),...

  - `priority_level_seat_count_watermarks` is removed.
  - `priority_level_request_count_samples` is replaced with `priority_level_request_utilization`, which samples every nanosecond rather than every millisecond; the old metric conveyed utilization despite its name.

* Fix watch to not send the same set of events multiple times causing watcher to go back in time ([#73845](https://github.com/kubernetes/kubernetes/pull/73845), [@wojtek-t](https://github.com/wojtek-t))

objects with a filter, when an update was made that no longer matched the filter a DELETE event was correctly sent.  However, the object that was returned by that delete was not the (correct) version before the update, but instead, the newer version.  That meant the new object was not matched by the filter.  This was a regression from behavior between cached watches on the server side and uncached watches, and thus broke downstream API clients. ([#46223](https://github.com/kubernetes/kubernetes/pull/46223),...