* Introduced API `tf.saved_model.experimental.read_fingerprint(export_dir)` for reading the fingerprint of a SavedModel.
* `tf.random`
  * Added non-experimental aliases for `tf.random.split` and `tf.random.fold_in`, the experimental endpoints are still available so no code changes are necessary.

- Check for newer resize fields when deciding recovery feature's status in kubelet ([#131437](https://github.com/kubernetes/kubernetes/pull/131437), [@gnufied](https://github.com/gnufied)) [SIG Storage]
- Disable reading of disk geometry before calling expansion for ext and xfs filesystems ([#131636](https://github.com/kubernetes/kubernetes/pull/131636), [@gnufied](https://github.com/gnufied)) [SIG Storage]

- The underlying format of the `CPUManager` state file has changed. Upgrades should be seamless, but any third-party tools that rely on reading the previous format need to be updated. ([#84462](https://github.com/kubernetes/kubernetes/pull/84462), [@klueska](https://github.com/klueska)) [SIG Node and Testing]

* Kubelet now exposes a new endpoint, `/metrics/probes`, which exposes a Prometheus metric containing the liveness and/or readiness probe results for a container. ([#61369](https://github.com/kubernetes/kubernetes/pull/61369), [@rramkumar1](https://github.com/rramkumar1))

### SIG Network

* On GCI, cleanup kubelet startup ([#35319](https://github.com/kubernetes/kubernetes/pull/35319), [@vishh](https://github.com/vishh))
* Add a retry when reading a file content from a container ([#35560](https://github.com/kubernetes/kubernetes/pull/35560), [@jingxu97](https://github.com/jingxu97))

- kubeadm: switched the kube-scheduler static Pod to use the endpoints `/livez` (for startup and liveness probes) and `/readyz` (for the readiness probe). Previously, `/healthz` was used for all probes, which is deprecated behavior in the scope of this component. ([#126945](https://github.com/kubernetes/kubernetes/pull/126945), [@liangyuanpeng](https://github.com/liangyuanpeng)) [SIG Cluster...

### Bug or Regression

- Fix 121094 by re-introducing the readiness predicate for externalTrafficPolicy: Local services. ([#121116](https://github.com/kubernetes/kubernetes/pull/121116), [@alexanderConstantinescu](https://github.com/alexanderConstantinescu)) [SIG Cloud Provider and Network]

			c.Lock()
			defer c.Unlock()
			c.state = srState{}
			c.enabled = false
		}
		return err
	}

	// attempt to read just the version key in the state file to ensure we
	// are reading a compatible version.
	var ver struct {
		Version int `json:"version"`
	}
	err = json.Unmarshal(buf, &ver)
	if err != nil {
		return err
	}
	if ver.Version != srStateFormatVersion1 {

### API Change

- 'ACTION REQUIRED: The Dynamic Resource Allocation (DRA) driver's DaemonSet
  must be deployed with a service account that enables writing ResourceSlice
  and reading ResourceClaim objects.'
   ([#125163](https://github.com/kubernetes/kubernetes/pull/125163), [@pohly](https://github.com/pohly)) [SIG Auth, Node and Testing]

* CoreDNS is now version 1.5.0 ([#78030](https://github.com/kubernetes/kubernetes/pull/78030), [@rajansandeep](https://github.com/rajansandeep))
    * A `ready` plugin has been included to report pod readiness
    * The `proxy` plugin has been deprecated. The `forward` plugin is to be used instead.
    * CoreDNS fixes the logging now that kubernetes’ client lib switched to klog from glog.