`autoscaling/v2alpha1`. The Horizontal Pod Autoscaler changes include support for status conditions, and autoscaling on memory and custom metrics.

* The metrics APIs, `custom-metrics.metrics.k8s.io` and `metrics`, were moved from `v1alpha1` to `v1beta1`, and renamed to `custom.metrics.k8s.io` and `metrics.k8s.io`, respectively. If you have deployed a custom metrics adapter, ensure that it supports the new API version. If you have deployed Heapster in aggregated API server mode, upgrade Heapster...

- If `ComponentSLIs` feature gate is enabled, then `/metrics/slis` becomes available on kubelet, allowing you to scrape health check metrics. ([#113030](https://github.com/kubernetes/kubernetes/pull/113030), [@Richabanker](https://github.com/Richabanker)) [SIG Node]

- Cloud controller manager's node controller now emits timing metrics for initial `Node` synchronization. These metrics measure the delay between the creation of a new `Node` and the node controller's initial management actions, such as removing the cloud provider taint. These metrics should be consulted when setting cloud controller manager's `--concurrent-node-syncs` flag. ([#119241](https://github.com/kubernetes/kubernetes/pull/119241),...

- Removes cAdvisor JSON metrics (/stats/container, /stats/<podname>/<containername>, /stats/<namespace>/<podname>/<poduid>/<containername>) from the kubelet. ([#99236](https://github.com/kubernetes/kubernetes/pull/99236), [@pacoxu](https://github.com/pacoxu))

- Improved misleading message, in case of no metrics received for the `HPA` controlled pods. ([#114740](https://github.com/kubernetes/kubernetes/pull/114740), [@kushagra98](https://github.com/kushagra98))
- Introduced new metrics removing the redundant subsystem in kube-apiserver pod logs metrics and deprecate the original ones:

- Metrics server nanny has now poll period set to 30s (previously 5 minutes) to allow faster scaling of metrics server. ([#101869](https://github.com/kubernetes/kubernetes/pull/101869), [@olagacek](https://github.com/olagacek)) [SIG Cloud Provider and Instrumentation]

- Added `apiserver_watch_list_duration_seconds` metrics which will measure response latency distribution in seconds for watchlist requests broken by group, version, resource and scope. ([#120490](https://github.com/kubernetes/kubernetes/pull/120490), [@p0lyn0mial](https://github.com/p0lyn0mial))

  - liveness/readiness probes to kube-scheduler MUST use HTTPS now, and the default port has been changed to 10259.
  - Applications that fetch metrics from kube-scheduler should use a dedicated service account which is allowed to access nonResourceURLs `/metrics`. ([#96345](https://github.com/kubernetes/kubernetes/pull/96345), [@ingvagabund](https://github.com/ingvagabund)) [SIG Cloud Provider, Scheduling and Testing]

- When metrics are counted, discard the wrong container StartTime metrics ([#110880](https://github.com/kubernetes/kubernetes/pull/110880), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Instrumentation and Node]

- Remove a v1alpha1 networking API for ClusterCIDRConfig ([#109436](https://github.com/kubernetes/kubernetes/pull/109436), [@JamesLaverack](https://github.com/JamesLaverack))
- Renamed metrics `evictions_number` to `evictions_total` and mark it as stable. The original `evictions_number` metrics name is marked as "Deprecated" and has been removed in kubernetes 1.23 . ([#106366](https://github.com/kubernetes/kubernetes/pull/106366), [@cyclinder](https://github.com/cyclinder))