CSI Migration enables the replacement of existing in-tree storage plugins such as `kubernetes.io/gce-pd` or `kubernetes.io/aws-ebs` with a corresponding CSI driver.
If CSI Migration is working properly, Kubernetes end users shouldn’t notice a difference.
After migration, Kubernetes users may continue to rely on all the functionality of in-tree storage plugins using the existing interface.

  then kube-proxy will send the traffic to the Terminating pod rather than dropping it. This
  patches up a race condition between when a pod is killed and when the external load
  balancer notices that it has been killed. ([#97238](https://github.com/kubernetes/kubernetes/pull/97238), [@andrewsykim](https://github.com/andrewsykim))

- Kubeadm: enabled the --experimental-watch-progress-notify-interval flag for etcd and set it to 5s. The flag specifies an interval at which etcd sends watch data to the kube-apiserver. ([#111383](https://github.com/kubernetes/kubernetes/pull/111383), [@p0lyn0mial](https://github.com/p0lyn0mial))

- Fixes a problem with 63-second or 1-second connection delays with some VXLAN-based
  network plugins which was first widely noticed in 1.16 (though some users saw it
  earlier than that, possibly only with specific network plugins). If you were previously
  using ethtool to disable checksum offload on your primary network interface, you should

  * You can remove the "old" taints (stored internally as annotations) at any time after the upgrade by doing `kubectl annotate nodes <node name> scheduler.alpha.kubernetes.io/taints-` (note the minus at the end, which means "delete the taint with this key")

consistency and security, Kubernetes performs pruning, or the automatic removal of unknown fields in objects sent to a Kubernetes API. An "unknown" field is one that is not specified in the OpenAPI validation schema. This behavior is already in place for native resources and ensures only data structures specified by the CRD developer are persisted to etcd. It will be available as a beta feature in Kubernetes 1.15.

Pruning is activated by setting `spec.preserveUnknownFields: false` in the Cu...

* The IP address and port for kube-proxy metrics server is now configurable via flag `--metrics-bind-address` ([#48625](https://github.com/kubernetes/kubernetes/pull/48625), [@mrhohn](https://github.com/mrhohn))
  * Special notice for kube-proxy in 1.7+ (including 1.7.0):
    * Healthz server (/healthz) will be served on 0.0.0.0:10256 by default.
    * Metrics server (/metrics and /proxyMode) will be served on 127.0.0.1:10249 by default.

- The scheduling queue didn't notice any extenders' failures, it could miss some cluster events,
  and it could end up Pods rejected by Extenders stuck in unschedulable pod pool in 5min in the worst-case scenario.