### Cluster Autoscaler
* If you are using (or planning to use) Cluster Autoscaler please wait for Kubernetes 1.6.1. In 1.6.0 Cluster Autoscaler

   ([#121461](https://github.com/kubernetes/kubernetes/pull/121461), [@alexzielenski](https://github.com/alexzielenski))
- Scheduler now skips the `NodeAffinity Score` plugin when it has nothing to do with a Pod. You might have noticed an increase in the metric `plugin_execution_duration_seconds` for `extension_point=score` and `plugin=NodeAffinity`, because the plugin only runs when it's relevant.

    - When upgrading from previous releases, issues will arise with public IPs if multiple clusters share the same resource group. To solve these problems, make the following changes to the cluster:
Recreate the relevant LoadBalancer services, or add a new tag 'kubernetes-cluster-name: <cluster-name>' manually for existing public IPs.

## Changelog since v1.3.5

### Other notable changes

* Addresses vSphere Volume Attach limits ([#29881](https://github.com/kubernetes/kubernetes/pull/29881), [@dagnello](https://github.com/dagnello))
* Increase request timeout based on termination grace period ([#31275](https://github.com/kubernetes/kubernetes/pull/31275), [@dims](https://github.com/dims))

* AWS: Fix long-standing bug in stringSetToPointers ([#26331](https://github.com/kubernetes/kubernetes/pull/26331), [@therc](https://github.com/therc))
* AWS kube-up: Increase timeout waiting for docker start ([#25405](https://github.com/kubernetes/kubernetes/pull/25405), [@justinsb](https://github.com/justinsb))

Required

* ACTION REQUIRED: The node.k8s.io API group and runtimeclasses.node.k8s.io resource have been migrated to a built-in API. If you were using RuntimeClasses (an default-disabled alpha feature, as of Kubernetes v1.12), then you must recreate all RuntimeClasses after upgrading, and the runtimeclasses.node.k8s.io CRD should be manually deleted. RuntimeClasses can no longer be created without a defined handler. ([#74433](https://github.com/kubernetes/kubernetes/pull/74433), [@tallclair...

* In GCE with COS, increase TasksMax for Docker service to raise cap on number of threads/processes used by containers. ([#51986](https://github.com/kubernetes/kubernetes/pull/51986), [@yujuhong](https://github.com/yujuhong))

- Number of errors reported to the metric `storage_operation_duration_seconds_count` for emptyDir decreased significantly because previously one error was reported for each projected volume created. ([#117022](https://github.com/kubernetes/kubernetes/pull/117022), [@mpatlasov](https://github.com/mpatlasov)) [SIG Storage]