- Fixed a regression in default 1.27 configurations in kube-apiserver: fixed the AggregatedDiscoveryEndpoint feature (beta in 1.27+) to successfully fetch discovery information from aggregated API servers that do not check `Accept` headers when serving the `/apis` endpoint ([#120360](https://github.com/kubernetes/kubernetes/pull/120360), [@Jefftree](https://github.com/Jefftree)) [SIG API Machinery]...

-  Forget rate limit when CRD establish controller successfully updated CRD condition ([#67370](https://github.com/kubernetes/kubernetes/pull/67370), [@yue9944882](https://github.com/yue9944882))

- Unhealthy pods covered by PDBs can be successfully evicted if enough healthy pods are available. ([#94381](https://github.com/kubernetes/kubernetes/pull/94381), [@michaelgugino](https://github.com/michaelgugino)) [SIG Apps]

* kubeadm upgrade no longer races leading to unexpected upgrade behavior on pod restarts ([#62655](https://github.com/kubernetes/kubernetes/pull/62655), [@stealthybox](https://github.com/stealthybox))
    * kubeadm upgrade now successfully upgrades etcd and the controlplane to use TLS
    * kubeadm upgrade now supports external etcd setups
    * kubeadm upgrade can now rollback and restore etcd after an upgrade failure

- Reverted graceful node shutdown to match 1.21 behavior of setting pods that have not yet successfully completed to "Failed" phase if the GracefulNodeShutdown feature is enabled in kubelet. The GracefulNodeShutdown feature is beta and must be explicitly configured via kubelet config to be enabled in 1.21+. This changes 1.22 and 1.23 behavior on...

* Skip podpreset test if the alpha feature settings/v1alpha1 is disabled ([#53080](https://github.com/kubernetes/kubernetes/pull/53080), [@jennybuckley](https://github.com/jennybuckley))
* Log when node is successfully initialized by Cloud Controller Manager ([#53517](https://github.com/kubernetes/kubernetes/pull/53517), [@andrewsykim](https://github.com/andrewsykim))

- Reverts graceful node shutdown to match 1.21 behavior of setting pods that have not yet successfully completed to "Failed" phase if the GracefulNodeShutdown feature is enabled in kubelet. The GracefulNodeShutdown feature is beta and must be explicitly configured via kubelet config to be enabled in 1.21+. This changes 1.22 and 1.23 behavior on...

- Reverts graceful node shutdown to match 1.21 behavior of setting pods that have not yet successfully completed to "Failed" phase if the GracefulNodeShutdown feature is enabled in kubelet. The GracefulNodeShutdown feature is beta and must be explicitly configured via kubelet config to be enabled in 1.21+. This changes 1.22 and 1.23 behavior on...

    *   A new flag `ignore_live_threads` is available on train. If set to
        `True`, it will ignore threads that remain running when tearing down
        infrastructure after successfully completing training, instead of
        throwing a RuntimeError.
    *   Restandardize `DenseVariational` as simpler template for other
        probabilistic layers.