- kubeadm: The kubeadm configuration now supports the definition of more than one control plane instances with their own APIEndpoint. The APIEndpoint for the "bootstrap" control plane instance should be defined using `InitConfiguration.APIEndpoint`, while the APIEndpoints for additional control plane instances should be added using `JoinConfiguration.APIEndpoint`. ([#67832](https://github.com/kubernetes/kubernetes/pull/67832), [@fabriziopandini](http...

- Deprecate the instance type beta label (`beta.kubernetes.io/instance-type`) in favor of its GA equivalent: `node.kubernetes.io/instance-type` ([#82049](https://github.com/kubernetes/kubernetes/pull/82049), [@andrewsykim](https://github.com/andrewsykim))

* Address a bug which allowed the horizontal pod autoscaler to allocate `desiredReplicas` > `maxReplicas` in certain instances. ([#53690](https://github.com/kubernetes/kubernetes/pull/53690), [@mattjmcnaughton](https://github.com/mattjmcnaughton))

 - Fix: support removal of nodes backed by deleted non VMSS instances on Azure ([#91184](https://github.com/kubernetes/kubernetes/pull/91184), [@bpineau](https://github.com/bpineau)) [SIG Cloud Provider]

  fix a kube-proxy skew policy misalignment. Its default value is `false`. Upgrade
  of the CoreDNS and kube-proxy addons will now trigger after all the control plane
  instances have been upgraded, unless the fearure gate is set to true. This feature
  gate will be removed in a future release. ([#117660](https://github.com/kubernetes/kubernetes/pull/117660), [@pacoxu](https://github.com/pacoxu))

- Avoid caching the Azure VMSS instances whose network profile is nil ([#100948](https://github.com/kubernetes/kubernetes/pull/100948), [@feiskyer](https://github.com/feiskyer)) [SIG Cloud Provider]

* Restores the pre-1.10 behavior of the openstack cloud provider which uses the instance name as the Kubernetes Node name. This requires instances be named with RFC-1123 compatible names. ([#63903](https://github.com/kubernetes/kubernetes/pull/63903), [@liggitt](https://github.com/liggitt))

*   Fixed a bug that allowed the horizontal pod autoscaler to allocate more `desiredReplica` objects than `maxReplica` objects in certain instances. ([#53690](https://github.com/kubernetes/kubernetes/pull/53690),[ @mattjmcnaughton](https://github.com/mattjmcnaughton))

### **AWS**

*   Nodes can now use instance types (such as C5) that use NVMe. ([#56607](https://github.com/kubernetes/kubernetes/pull/56607), [@justinsb](https://github.com/justinsb))

- Fixed a regression in the pod binding subresource to honor the `metadata.uid` precondition.
  This allows kube-scheduler to ensure it is assigns node names to the same instances of pods it made scheduling decisions for. ([#116550](https://github.com/kubernetes/kubernetes/pull/116550), [@alculquicondor](https://github.com/alculquicondor))

* Restored the pre-1.10 behavior of the openstack cloud provider, which uses the instance name as the Kubernetes Node name. This requires instances be named with RFC-1123 compatible names. ([#63903](https://github.com/kubernetes/kubernetes/pull/63903), [@liggitt](https://github.com/liggitt))