- `kube-proxy` with `--proxy-mode=ipvs` can be used with statically linked kernels.
  The reseved IPv4 range `TEST-NET-2` in `rfc5737` MUST NOT be used for `ClusterIP` or `loadBalancerIP` since address `198.51.100.0` is used for probing. ([#114669](https://github.com/kubernetes/kubernetes/pull/114669), [@uablrek](https://github.com/uablrek))

- When the `PodObservedGenerationTracking` feature gate was set, the kubelet populated `status.observedGeneration` to reflect the latest `metadata.generation` it observed for the pod. ([#130352](https://github.com/kubernetes/kubernetes/pull/130352), [@natasha41575](https://github.com/natasha41575)) [SIG API Machinery, Apps, CLI, Node, Release, Scheduling, Storage, Testing and Windows]

        *   Eager performance should be similar with this feature enabled.
            *   A roughly 5us per-op overhead may be observed when running many
                small functions.
            *   Note a
                [known issue](https://github.com/tensorflow/tensorflow/issues/55414)
                with GPU performance.

- CRDs: For structural schemas, non-nullable null map fields will now be dropped and defaulted if a default is available. null items in the list will continue being preserved, and fail validation if not nullable. ([#95423](https://github.com/kubernetes/kubernetes/pull/95423), [@apelisse](https://github.com/apelisse)) [SIG API Machinery]

* Pass cluster_location argument to Heapster ([#65176](https://github.com/kubernetes/kubernetes/pull/65176), [@kawych](https://github.com/kawych))
* fixes a memory leak in the kube-controller-manager observed when large numbers of pods with tolerations are created/deleted ([#65339](https://github.com/kubernetes/kubernetes/pull/65339), [@liggitt](https://github.com/liggitt))