* Role, ClusterRole and their bindings for cloud-provider is put under system namespace. Their addonmanager mode switches to EnsureExists. ([#67224](https://github.com/kubernetes/kubernetes/pull/67224), [@grayluck](https://github.com/grayluck))

### Bug or Regression

- Fix scoring for NodeResourcesMostAllocated and NodeResourcesBalancedAllocation plugins when nodes have containers with no requests. This was leaving to under-utilization of small nodes. ([#102925](https://github.com/kubernetes/kubernetes/pull/102925), [@alculquicondor](https://github.com/alculquicondor)) [SIG Scheduling]

* Ensure cloudprovider.InstanceNotFound is reported when the VM is not found on Azure ([#61531](https://github.com/kubernetes/kubernetes/pull/61531), [@feiskyer](https://github.com/feiskyer))
* On AWS kubelet returns an error when started under conditions that do not allow it to work (AWS has not yet tagged the instance). ([#60125](https://github.com/kubernetes/kubernetes/pull/60125), [@vainu-arto](https://github.com/vainu-arto))

  - https://github.com/kubernetes-sigs/kustomize/issues/4731: Remove a potential build-time side-effect in ConfigMapGenerator and SecretGenerator, which loaded values from the local environment under some circumstances, breaking kustomize build's side-effect-free promise. While this behavior was never intended, we deprecated it and are announcing it as a breaking change since it existed for a long time. See also the Eschewed Features documentation....

- the following deprecated APIs can no longer be served:
  - All resources under `apps/v1beta1` and `apps/v1beta2` - use `apps/v1` instead
  - `daemonsets`, `deployments`, `replicasets` resources under `extensions/v1beta1` - use `apps/v1` instead
  - `networkpolicies` resources under `extensions/v1beta1` - use `networking.k8s.io/v1` instead

- 'KubeSchedulerConfiguration' which used to be under GroupVersion 'componentconfig/v1alpha1',
is now under 'kubescheduler.config.k8s.io/v1alpha1'.  ([#66916](https://github.com/kubernetes/kubernetes/pull/66916), [@dixudx](https://github.com/dixudx)) Courtesy of SIG Cluster Lifecycle, SIG Scheduling, and SIG Testing

* On AWS kubelet returns an error when started under conditions that do not allow it to work (AWS has not yet tagged the instance). ([#60125](https://github.com/kubernetes/kubernetes/pull/60125), [@vainu-arto](https://github.com/vainu-arto))

 - Scheduling failures due to no nodes available are now reported as unschedulable under ```schedule_attempts_total``` metric. ([#90989](https://github.com/kubernetes/kubernetes/pull/90989), [@ahg-g](https://github.com/ahg-g)) [SIG Scheduling]

### Other (Cleanup or Flake)

* Get public IP for Azure vmss nodes. ([#68498](https://github.com/kubernetes/kubernetes/pull/68498), [@feiskyer](https://github.com/feiskyer))
* Role, ClusterRole and their bindings for cloud-provider is put under system namespace. Their addonmanager mode switches to EnsureExists. ([#67224](https://github.com/kubernetes/kubernetes/pull/67224), [@grayluck](https://github.com/grayluck))

- `kubeadm init` and `kubeadm init phase` now have a `--image-repository` flag, improving support for environments with limited access to official kubernetes repository.
- The DynamicKubeletConfig and SelfHosting functionality was moved outside of `kubeadm init` and feature gates and is now exposed under `kubeadm alpha`.