- Fixed a bug in the single-numa-policy of the TopologyManager. Previously, best-effort pods would result in a terminated state with a TopologyAffinity error. Now they will run as expected. ([#83777](https://github.com/kubernetes/kubernetes/pull/83777), [@lmdaly](https://github.com/lmdaly))

  After the connection closed(received GOAWAY), the client's other in-flight requests won't be affected, and the client will reconnect. 
  The flag min value is 0 (off), max is .02 (1/50 requests); .001 (1/1000) is a recommended starting point.

* Fix kubelet to correctly umounts mount points for glusterfs when transport endpoint is not connected and nfs when there is a stale file handle ([#58660](https://github.com/kubernetes/kubernetes/pull/58660), [@humblec](https://github.com/humblec))
* Detach and clear bad disk URI ([#58345](https://github.com/kubernetes/kubernetes/pull/58345), [@rootfs](https://github.com/rootfs))

### Other notable changes

- Fixed a regression in kube-proxy where it might refuse to start if given
  single-stack IPv6 configuration options on a node that has both IPv4 and
  IPv6 IPs. ([#121008](https://github.com/kubernetes/kubernetes/pull/121008), [@danwinship](https://github.com/danwinship)) [SIG Network]

# code.
#
# https://github.com/kubernetes/test-infra/blob/master/prow/plugins/size/size.go
#
# This file is a series of lines, each of the form:
#     <type> <name>
#
# Type can be:
#    path - an exact path to a single file
#    file-name - an exact leaf filename, regardless of path
#    path-prefix - a prefix match on the file path
#    file-prefix - a prefix match of the leaf filename (no path)

* Added anti-affinity to kube-dns pods. Otherwise the "no single point of failure" setting doesn't actually work (a single node failure can still take down the entire cluster). ([#57683](https://github.com/kubernetes/kubernetes/pull/57683), [@vainu-arto](https://github.com/vainu-arto))

### API-Machinery

    *  - vSphere cloud provider (VCP) uses OS hostname and not vSphere Inventory VM Name.
    *    That means, now  VCP can handle cases where user changes VM inventory name.
    * - VCP can handle cases where VM migrates to other ESXi cluster or datacenter or vCenter.
    * The only requirement is the shared storage. VCP needs shared storage on all Node VMs.

- Fixes a rare race condition handling requests that timeout ([#107460](https://github.com/kubernetes/kubernetes/pull/107460), [@liggitt](https://github.com/liggitt)) [SIG API Machinery]
- Mount-utils: Detect potential stale file handle ([#107040](https://github.com/kubernetes/kubernetes/pull/107040), [@andyzhangx](https://github.com/andyzhangx)) [SIG Storage]

### Other (Cleanup or Flake)

- Fixed an EndpointSlice Controller hashing bug that could cause EndpointSlices to incorrectly handle Pods with duplicate IP addresses. For example this could happen when a new Pod reused an IP that was also assigned to a Pod in a completed state. ([#115907](https://github.com/kubernetes/kubernetes/pull/115907), [@qinqon](https://github.com/qinqon))...