- Fixes a bug where Services using finalizers may hold onto ClusterIP and/or NodePort allocated resources for longer than expected if the finalizer is removed using the status subresource ([#120655](https://github.com/kubernetes/kubernetes/pull/120655), [@aojea](https://github.com/aojea)) [SIG Network and Testing]...

- When using `kubectl replace` (or the equivalent API call) on a Service, the caller no longer needs to do a read-modify-write cycle to fetch the allocated values for `.spec.clusterIP` and `.spec.ports[].nodePort`.  Instead the API server will automatically carry these forward from the original object when the new object does not specify them. ([#104673](https://github.com/kubernetes/kubernetes/pull/104673), [@thockin](https:...

    *   vSphere: Fix disk is not getting detached when PV is provisioned on clustered datastore. ([#54438](https://github.com/kubernetes/kubernetes/pull/54438),[ @pshahzeb](https://github.com/pshahzeb))

* A cluster using the AWS cloud provider will need to label existing nodes and resources with a ClusterID or the kube-controller-manager will not start.  To run without a ClusterID pass --allow-untagged-cloud=true to the kube-controller-manager on startup. ([#49215](https://github.com/kubernetes/kubernetes/pull/49215), [@rrati](https://github.com/rrati))

* Allow access to ClusterIP from the host network namespace when kube-proxy is started in IPVS mode without either masqueradeAll or clusterCIDR flags ([#65388](https://github.com/kubernetes/kubernetes/pull/65388), [@lbernail](https://github.com/lbernail))

- Added optional `--detect-local-mode` flag to kube-proxy. Valid values are "ClusterCIDR" (default matching previous behavior) and "NodeCIDR" ([#87748](https://github.com/kubernetes/kubernetes/pull/87748), [@satyasm](https://github.com/satyasm)) [SIG Cluster Lifecycle, Network and Scheduling]

- Fixes a bug where Services using finalizers may hold onto ClusterIP and/or NodePort allocated resources for longer than expected if the finalizer is removed using the status subresource ([#120654](https://github.com/kubernetes/kubernetes/pull/120654), [@aojea](https://github.com/aojea)) [SIG Testing]

* Allow access to ClusterIP from the host network namespace when kube-proxy is started in IPVS mode without either masqueradeAll or clusterCIDR flags ([#65388](https://github.com/kubernetes/kubernetes/pull/65388), [@lbernail](https://github.com/lbernail))

- Services will now have a `clusterIPs` field to go with `clusterIP`.  `clusterIPs[0]` is a synonym for `clusterIP` and will be synchronized on create and update operations. ([#95894]...

    deletion of the Node object, deletion of the instance in the cloud provider,
    or force deletion of the pod from the api-server).
    This helps prevent "split brain" scenarios in clustered applications by
    ensuring that unreachable pods will not be presumed dead unless some
    "fencing" operation has provided one of the above indications.