- Updates that remove remaining `metadata.finalizers` from  an object that is pending deletion (non-nil metadata.deletionTimestamp) and has no graceful deletion pending (nil or 0 metadata.deletionGracePeriodSeconds) now results in immediate deletion of the object. ([#77952](https://github.com/kubernetes/kubernetes/pull/77952), [@liggitt](https://github.com/liggitt))

// and not a part of non-test builds.
type http2synctestGroupInterface interface {
	Join()
	Now() time.Time
	NewTimer(d time.Duration) http2timer
	AfterFunc(d time.Duration, f func()) http2timer
	ContextWithTimeout(ctx context.Context, d time.Duration) (context.Context, context.CancelFunc)
}

// pipe is a goroutine-safe io.Reader/io.Writer pair. It's like

   ([#122830](https://github.com/kubernetes/kubernetes/pull/122830), [@p0lyn0mial](https://github.com/p0lyn0mial))
- Updated `ImageGCMaxAge` behavior in the kubelet to wait the `MaxAge` duration after the kubelet has restarted before garbage collecting.
   ([#123343](https://github.com/kubernetes/kubernetes/pull/123343), [@haircommander](https://github.com/haircommander))

* fix azure file plugin failure issue on Windows after node restart ([#60625](https://github.com/kubernetes/kubernetes/pull/60625), [@andyzhangx](https://github.com/andyzhangx))
* Fixes the races around devicemanager Allocate() and endpoint deletion. ([#60856](https://github.com/kubernetes/kubernetes/pull/60856), [@jiayingz](https://github.com/jiayingz))



# v1.9.6

  // CHECK-SAME: batch_group_count = 1
  // CHECK-SAME: feature_group_count = 2

  %0 = "tf.Conv2D"(%arg0, %arg1) {data_format = "NHWC", dilations = [1, 2, 3, 1], padding = "SAME", strides = [1, 4, 5, 1]} : (tensor<256x32x32x6xf32>, tensor<3x3x3x16xf32>) -> tensor<256x8x7x16xf32>
  func.return %0 : tensor<256x8x7x16xf32>
}

// -----

- The `controller.kubernetes.io/pod-deletion-cost` annotation can be set to offer a hint on the cost of deleting a `Pod` compared to other pods belonging to the same ReplicaSet. Pods with lower deletion cost are deleted first. This is an alpha feature. ([#99163](https://github.com/kubernetes/kubernetes/pull/99163), [@ahg-g](https://github.com/ahg-g))

* Fixed a bug in the apiserver storage that could cause just-added finalizers to be ignored on an immediately following delete request, leading to premature deletion. ([#77619](https://github.com/kubernetes/kubernetes/pull/77619), [@caesarxuchao](https://github.com/caesarxuchao))

- Adds `\livez` for liveness health checking for kube-apiserver. Using the parameter `--maximum-startup-sequence-duration` will allow the liveness endpoint to defer boot-sequence failures for the specified duration period. ([#81969](https://github.com/kubernetes/kubernetes/pull/81969), [@logicalhan](https://github.com/logicalhan))

  without risking breaking old clients, such as controllers.

- `kubectl delete` no longer scales down workload API objects before deletion.
  Users who depend on ordered termination for the Pods of their StatefulSets
  must use `kubectl scale` to scale down the StatefulSet before deletion.

- `kubectl run --env` no longer supports CSV parsing. To provide multiple environment

## Changelog since v1.7.14

### Other notable changes

* fixed foreground deletion of podtemplates ([#60683](https://github.com/kubernetes/kubernetes/pull/60683), [@nilebox](https://github.com/nilebox))