- Fixed two scheduler metrics...

* Add sequential allocator for device names. ([#38818](https://github.com/kubernetes/kubernetes/pull/38818), [@jsafrane](https://github.com/jsafrane))

#### Azure
* Fix failing load balancers in Azure ([#40405](https://github.com/kubernetes/kubernetes/pull/40405), [@codablock](https://github.com/codablock))

    * randomly select which nodes are members of Kubernetes External Load Balancers.
* Clarify supported version skew between masters, nodes, and clients ([#25087](https://github.com/kubernetes/kubernetes/pull/25087), [@ihmccreery](https://github.com/ihmccreery))

#### Load Balancing
* [stable] Source IP Preservation - change Cloud load-balancer strategy to health-checks and respond to health check only on nodes that host pods for the service. See [Create an External Load Balancer - Preserving the client source IP](https://kubernetes.io/docs/tasks/access-application-cluster/create-external-load-balancer/#preserving-the-client-source-ip).
  Two annotations have been promoted to API fields:

- Fixed a regression introduced in 1.24.0 where Azure load balancers were not kept up to date with the state of cluster nodes. In particular, nodes that are not in the ready state and are not newly created (i.e. not having the `node.cloudprovider.kubernetes.io/uninitialized` taint) now get removed from Azure load balancers. ([#109931](https://github.com/kubernetes/kubernetes/pull/109931), [@ricky-rav](https://github.com/ricky-rav)) [SIG Cloud...

providers now waits for the providerID value to be available before untainting it. This ensures that nodes are not declared Ready without necessary information such as the providerID and zone labels, which are required for integrations like load balancers to function correctly. Cloud providers that do not implement the GetInstanceProviderID method will not require the providerID to be set and will not fail to initialize the node for backward compatibility. ([#123713](https://github.com/kubernetes...

unauthenticated. The mitigation may be enabled by setting the `UnauthenticatedHTTP2DOSMitigation` feature gate to `true` (it is disabled by default). An API server fronted by an L7 load balancer that already mitigates these http/2 attacks may choose not to enable the kube-apiserver mitigation to avoid disrupting load balancer → kube-apiserver connections if http/2 requests from multiple clients share the same backend connection. An API server on a private network may choose not to enable the kube-apiserver...

- Service load balancers no longer exclude nodes marked unschedulable from the candidate nodes. The service load balancer exclusion label should be used instead.
  

  - the node label `alpha.service-controller.kubernetes.io/exclude-balancer` which excludes a node from cloud load balancers (using Service Type=LoadBalancer) is deprecated in favor of `node.kubernetes.io/exclude-balancer`. Support for `alpha.service-controller.kubernetes.io/exclude-balancer` will be removed in v1.18.

- Features:

unauthenticated. The mitigation may be disabled by setting the `UnauthenticatedHTTP2DOSMitigation` feature gate to `false` (it is enabled by default). An API server fronted by an L7 load balancer that already mitigates these http/2 attacks may choose to disable the kube-apiserver mitigation to avoid disrupting load balancer -> kube-apiserver connections if http/2 requests from multiple clients share the same backend connection. An API server on a private network may opt to disable the kube-apiserver mitigation...