// The scheduler will prefer to schedule pods to nodes that satisfy
  // the affinity expressions specified by this field, but it may choose
  // a node that violates one or more of the expressions. The node that is
  // most preferred is the one with the greatest sum of weights, i.e.
  // for each node that meets all of the scheduling requirements (resource

- `kubeadm`: will now allow deploying a kubelet that is 3 versions older than the version of `kubeadm` (N-3). This aligns with the recent change made by SIG Architecture that extends the support skew between the control plane and kubelets. Tolerate this new kubelet skew for the commands `init`, `join` and `upgrade`. Note that if the `kubeadm` user applies a control plane version that is older than the `kubeadm` version (N-1 maximum) then the skew between the kubelet and...

- A terminating pod on a node that is not caused by preemption no longer prevents `kube-scheduler` from preempting pods on that node

### Other notable changes

* Added new flag to `kubeadm init`: --node-name, that lets you specify the name of the Node object that will be created ([#48594](https://github.com/kubernetes/kubernetes/pull/48594), [@GheRivero](https://github.com/GheRivero))

Traditionally, the Kubernetes scheduler was based on the assumptions that additional persistent storage is available everywhere in the cluster and has infinite capacity. Topology constraints addressed the first point, but up to now pod scheduling was still done without considering that the remaining storage capacity may not be enough to start a new pod. [Storage capacity tracking](https://docs.k8s.io/concepts/storage/storage-capacity/) addresses that by adding an API for a CSI driver to report storage capacity...

*   A new controller automatically cleans up Certificate Signing Requests that are Approved and Issued, or Denied. ([#51840](https://github.com/kubernetes...

external etcd endpoints passed in the `ClusterConfiguration` that have Unicode characters are no longer IDNA encoded (converted to Punycode). They are now just URL encoded as per Go's implementation of RFC-3986, have duplicate "/" removed from the URL paths, and passed like that directly to the `kube-apiserver` `--etcd-servers` flag. If you have etcd endpoints that have Unicode characters, it is advisable to encode them in advance with tooling that is fully IDNA compliant. If you don't do that, the Go standard...

    - [<code>CSIMigrationvSphere</code> feature gate has not migrated to new CRD APIs](#csimigrationvsphere-feature-gate-has-not-migrated-to-new-crd-apis)
    - [Workloads that saturate nodes with pods may see pods that fail due to node admission](#workloads-that-saturate-nodes-with-pods-may-see-pods-that-fail-due-to-node-admission)
    - [Etcd v3.5.[0-2] data corruption](#etcd-v350-2-data-corruption)
  - [Urgent Upgrade Notes](#urgent-upgrade-notes)

- UDP and SCTP protocols can left stale connections that need to be cleared to avoid services disruption, but they can cause problems that are hard to debug.

- e2e tests that require SSH may now be used against clusters that have nodes without external IP addresses by setting the environment variable `KUBE_SSH_BASTION` to the `host:port` of a machine that is allowed to SSH to those nodes.  The same private key that the test would use is used for the bastion host.  The test connects to the bastion and then tunnels another SSH connection to...