- GCE Internal LoadBalancer sync loop will now release the ILB IP address upon sync failure. An error in ILB forwarding rule creation will no longer leak IP addresses. ([#97740](https://github.com/kubernetes/kubernetes/pull/97740), [@prameshj](https://github.com/prameshj)) [SIG Cloud Provider and Network]

## Dependencies

- Pods will now post their readiness during termination. ([#110416](https://github.com/kubernetes/kubernetes/pull/110416), [@aojea](https://github.com/aojea)) [SIG Network, Node and Testing]

- The `/debug/api_priority_and_fairness/dump_requests` path at an apiserver will no longer return a phantom line for each exempt priority level. ([#93406](https://github.com/kubernetes/kubernetes/pull/93406), [@MikeSpreitzer](https://github.com/MikeSpreitzer)) [SIG API Machinery]

    * => IPv6 BackendPool name == <clusterName>-IPv6
    * This result into:
    * - clusters moving from IPv4 to duakstack will require no changes
    * - clusters moving from IPv6 (while not seen in the wild, we can not rule out their existence) to dualstack will require deleting backend pools (the reconciler will take care of creating correct backendpools)

* If ExperimentalCriticalPodAnnotation=True flag gate is set, kubelet will ensure that pods with `scheduler.alpha.kubernetes.io/critical-pod` annotation will be admitted even under resource pressure, will not be evicted, and are reasonably protected from system OOMs. ([#41052](https://github.com/kubernetes/kubernetes/pull/41052), [@vishh](https://github.com/vishh))

- Pods will now post their readiness during termination. ([#110418](https://github.com/kubernetes/kubernetes/pull/110418), [@aojea](https://github.com/aojea)) [SIG Network, Node and Testing]

- Pods will now post their readiness during termination. ([#110417](https://github.com/kubernetes/kubernetes/pull/110417), [@aojea](https://github.com/aojea)) [SIG Network, Node and Testing]

### Other (Cleanup or Flake)

- A v2-level info log will be added, which will output the details of the pod being preempted, including victim and preemptor ([#117214](https://github.com/kubernetes/kubernetes/pull/117214), [@HirazawaUi](https://github.com/HirazawaUi)) [SIG Scheduling]

## Dependencies

In Kubernetes 1.19 this feature will be enabled by default with kube-proxy reading from EndpointSlices instead of Endpoints. Although this will mostly be an invisible change, it should result in noticeable scalability improvements in large clusters. It will also enable significant new features in future Kubernetes releases like Topology Aware Routing.

- Introducing new semantic for metric `volume_operation_total_seconds` to be the end to end latency of volume provisioning/deletion. Existing metric "storage_operation_duration_seconds" will remain untouched, however it is exposed to the following potential...