- The unreachable taint gets applied to a node when it loses its network connection. ([#67734](https://github.com/kubernetes/kubernetes/pull/67734), [@Huang-Wei](https://github.com/Huang-Wei))

* kube-apiserver: fixes error creating system priority classes when starting multiple apiservers simultaneously ([#67372](https://github.com/kubernetes/kubernetes/pull/67372), [@tanshanshan](https://github.com/tanshanshan))

  - `requests.deviceclass.resource.k8s.io/<deviceclass>` is charged based on the worst-case number of devices requested.
  - Device classes mapping to an extended resource now consume `requests.<extended resource name>`. ([#134210](https://github.com/kubernetes/kubernetes/pull/134210), [@yliaog](https://github.com/yliaog)) [SIG API Machinery, Apps, Node, Scheduling and Testing]

- Fixed cluster autoscaler addon permissions so it can access batch/job. ([#69858](https://github.com/kubernetes/kubernetes/pull/69858), [@losipiuk](https://github.com/losipiuk))

### Bug or Regression

- Fix a scheduler preemption issue where the victim pod was not deleted due to incorrect status patching. This issue occurred when the preemptor and victim pods had different QoS classes in their status, causing the preemption to fail entirely. ([#126693](https://github.com/kubernetes/kubernetes/pull/126693), [@Huang-Wei](https://github.com/Huang-Wei)) [SIG Scheduling]

- The topology manager aligns resources for pods of all QoS classes with respect to NUMA locality, not just Guaranteed QoS pods. ([#83492](https://github.com/kubernetes/kubernetes/pull/83492), [@ConnorDoyle](https://github.com/ConnorDoyle))

SubjectAccessReview API in the `authorization.k8s.io` API group now allows providing the user uid. ([#49677](https://github.com/kubernetes/kubernetes/pull/49677), [@dims](https://github.com/dims))

* After a kubelet rotates its client cert, it now closes its connections to the API server to force a handshake using the new cert. Previously, the kubelet could keep its existing connection open, even if the cert used for that connection was expired and rejected by the API server. ([#49899](https://gi...

* Scheduler should terminate when it looses leader lock. ([#81306](https://github.com/kubernetes/kubernetes/pull/81306), [@ravisantoshgudimetla](https://github.com/ravisantoshgudimetla))

- Removed "endpoint-controller" and "workload-leader-election" FlowSchemas from the default APF configuration.
  
  migrate the lock type used in the leader election in your workloads from configmapsleases/endpointsleases to leases. ([#131215](https://github.com/kubernetes/kubernetes/pull/131215), [@tosi3k](https://github.com/tosi3k)) [SIG API Machinery, Apps, Network, Scalability and Scheduling]

* Disallow PriorityClass names with 'system-' prefix for user defined priority classes. ([#59382](https://github.com/kubernetes/kubernetes/pull/59382), [@bsalamat](https://github.com/bsalamat))