- Added new option to the `InterPodAffinity` scheduler plugin to ignore existing
  pods` preferred inter-pod affinities if the incoming pod has no preferred inter-pod
  affinities. This option can be used as an optimization for higher scheduling throughput
  (at the cost of an occasional pod being scheduled non-optimally/violating existing
  pods preferred inter-pod affinities). To enable this scheduler option, set the

nodes by available storage capacity. It's in alpha and disabled by default. The `VolumeCapacityPriority` alpha feature was replaced with this, and the default behavior was changed. The `VolumeCapacityPriority` preferred a node with the least allocatable, but the `StorageCapacityScoring` preferred a node with the maximum allocatable. See [KEP-4049](https://github.com/kubernetes/enhancements/blob/master/keps/sig-storage/4049-storage-capacity-scoring-of-nodes-for-dynamic-provisioning/README.md) for details....

* Job controller: Retry with rate-limiting if a Pod create/delete operation fails. ([#45870](https://github.com/kubernetes/kubernetes/pull/45870), [@tacy](https://github.com/tacy))
* Ensure that autoscaling/v1 is the preferred version for API discovery when autoscaling/v2alpha1 is enabled. ([#45741](https://github.com/kubernetes/kubernetes/pull/45741), [@DirectXMan12](https://github.com/DirectXMan12))

### Other notable changes

* [GCE] Apiserver uses `InternalIP` as the most preferred kubelet address type by default. ([#59019](https://github.com/kubernetes/kubernetes/pull/59019), [@MrHohn](https://github.com/MrHohn))

### Bug or Regression

- A node IP provided to kublet via `--node-ip` will now be preferred for when determining the node's primary IP and using the external cloud provider (CCM). ([#107750](https://github.com/kubernetes/kubernetes/pull/107750), [@stephenfin](https://github.com/stephenfin))

- Achieved 2X performance improvement on both required and preferred PodAffinity. ([#76243](https://github.com/kubernetes/kubernetes/pull/76243), [@Huang-Wei](https://github.com/Huang-Wei))

- The Image Locality priority function of the scheduler has been improved and is now enabled by default. With this feature enabled, nodes that have all or a partial set of images required for running a pod are preferred over other nodes, which improves pod start-up time.
- TaintNodeByCondition has been moved to Beta and is enabled by default.
- Scheduler throughput has been improved by ~50% in large clusters (>2000 nodes).

    * nonexistent.  This patch adds an anonymous function call inside
    * the loop to correctly handle acquiring the write lock with deferred
    * unlock to run at that function's completion, following the normal
    * golang lock/deferred-unlock pattern.
* fix UnmountDevice failure on Windows ([#68608](https://github.com/kubernetes/kubernetes/pull/68608), [@andyzhangx](https://github.com/andyzhangx))

    * nonexistent.  This patch adds an anonymous function call inside
    * the loop to correctly handle acquiring the write lock with deferred
    * unlock to run at that function's completion, following the normal
    * golang lock/deferred-unlock pattern.

### Windows