- Kube-apiserver: Each unique set of etcd server overrides specified with `--etcd-servers-overrides` surfaced health checks named `etcd-override-<index>` and `etcd-override-readiness-<index>`. These checks were still excluded by the `?exclude=etcd` and `?exclude=etcd-readiness` directives. ([#129438](https://github.com/kubernetes/kubernetes/pull/129438), [@pacoxu](https://github.com/pacoxu)) [SIG API Machinery and Testing]

- Azure cloud provider now supports unmanaged nodes (such as on-prem) that are labeled with `kubernetes.azure.com/managed=false` and `alpha.service-controller.kubernetes.io/exclude-balancer=true` ([#67984](https://github.com/kubernetes/kubernetes/pull/67984), [@feiskyer](https://github.com/feiskyer)) Courtesy of SIG Azure, and SIG Cloud Provider

* Fix problems with scaling up the cluster when unschedulable pods have some persistent volume claims. ([#44860](https://github.com/kubernetes/kubernetes/pull/44860), [@mwielgus](https://github.com/mwielgus))
* Exclude nodes labeled as master from LoadBalancer / NodePort; restores documented behaviour. ([#44745](https://github.com/kubernetes/kubernetes/pull/44745), [@justinsb](https://github.com/justinsb))

*   The "ServiceNodeExclusion" feature gate must now be enabled for the `alpha.service-controller.kubernetes.io/exclude-balancer` annotation on nodes to be honored. ([#54644](https://github.com/kubernetes/kubernetes/pull/54644),[ @brendandburns](https://github.com/brendandburns))

### **Scheduling**

- Revised the logic for DaemonSet rolling update to exclude nodes if scheduling constraints are not met.

  - 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:

- The `--cidr-allocator-type` option set to `CloudAllocator` for `kube-controller-manager` will be deprecated and removed in a future release. Users are advised to transition to and explore the available options provided by their external cloud provider. ([#123011](https://github.com/kubernetes/kubernetes/pull/123011), [@dims](https://github.com/dims))

- Add latency and request count metrics for scheduler framework. ([#83569](https://github.com/kubernetes/kubernetes/pull/83569), [@liu-cong](https://github.com/liu-cong))
- Expose SharedInformerFactory in the framework handle ([#83663](https://github.com/kubernetes/kubernetes/pull/83663), [@draveness](https://github.com/draveness))

- Revised the logic for `DaemonSet` rolling update to exclude nodes if scheduling constraints are not met. This eliminates the problem of rolling updates to a `DaemonSet` getting stuck around tolerations. ([#119317](https://github.com/kubernetes/kubernetes/pull/119317), [@mochizuki875](https://github.com/mochizuki875))...

- Fix a regression where the Kubelet failed to exclude already completed pods from calculations about how many resources it was currently using when deciding whether to allow more pods. ([#104577](https://github.com/kubernetes/kubernetes/pull/104577), [@smarterclayton](https://github.com/smarterclayton))...