Specifically, we will now reduce the amount of syncs for `ETP=local` services even
  further in the CCM and avoid re-configuring LBs to an even greater extent. ([#111658](https://github.com/kubernetes/kubernetes/pull/111658), [@alexanderConstantinescu](https://github.com/alexanderConstantinescu))

- Extend the Job API for alpha version of BackoffLimitPerIndex ([#119294](https://github.com/kubernetes/kubernetes/pull/119294), [@mimowo](https://github.com/mimowo)) [SIG API Machinery and Apps]

- Some interfaces' signatures in the scheduler were updated: - PluginsRunner: used NodeInfo in `RunPreScorePlugins` and `RunScorePlugins`. - PreScorePlugin: used NodeInfo in `PreScore`. - Extender: used NodeInfo in `Filter` and `Prioritize`.
   ([#121954](https://github.com/kubernetes/kubernetes/pull/121954), [@AxeZhan](https://github.com/AxeZhan))

- Exposed health check SLI metrics on `metrics/slis` for apiserver. ([#112741](https://github.com/kubernetes/kubernetes/pull/112741), [@logicalhan](https://github.com/logicalhan))
- Extend the job `job_finished_total metric by new `reason` label and introduce a new job metric to count pod failures

- `kubeadm`: will now allow deploying a kubelet that is 3 versions older than the version of `kubeadm` (N-3). This aligns with the recent change made by SIG Architecture that extends the support skew between the control plane and kubelets. Tolerate this new kubelet skew for the commands `init`, `join` and `upgrade`. Note that if the `kubeadm` user applies a control plane version that is older than the `kubeadm` version (N-1 maximum)...