#### Garbage Collector
* The garbage collector now supports custom APIs added via Custom Resource Definitions
  or aggregated API servers. The garbage collector controller refreshes periodically.
  Therefore, expect a latency of about 30 seconds between when an API is added and when
  the garbage collector starts to manage it.


#### Monitoring/Prometheus

- Fixed a bug in the garbage collector controller which could block indefinitely due to a cache sync failure. This fix allows the garbage collector to eventually continue garbage collecting other resources if a given resource cannot be listed or watched. Any objects in the unsynced resource type with owner references with `blockOwnerDeletion: true` will not be known to the garbage collector. Use of `blockOwnerDeletion` has always been best-effort and racy on...

* fix data loss issue if using existing azure disk with partitions in disk mount  ([#63270](https://github.com/kubernetes/kubernetes/pull/63270), [@andyzhangx](https://github.com/andyzhangx))
* The garbage collector now supports CustomResourceDefinitions and APIServices. ([#65916](https://github.com/kubernetes/kubernetes/pull/65916), [@xmudrii](https://github.com/xmudrii))

### Garbage Collector

* Finalizers are now honored on custom resources, and on other resources even when garbage collection is disabled via the apiserver flag `--enable-garbage-collector=false` ([#51148](https://github.com/kubernetes/kubernetes/pull/51148), [@ironcladlou](https://github.com/ironcladlou))

* Finalizers are now honored on custom resources, and on other resources even when garbage collection is disabled via the apiserver flag `--enable-garbage-collector=false` ([#51469](https://github.com/kubernetes/kubernetes/pull/51469), [@ironcladlou](https://github.com/ironcladlou))

- Fixed a regression in default configurations, which enabled PodDisruptionConditions by default, 

  - A cluster-scoped object with an ownerReference referencing a uid of a namespaced kind is now consistently treated as though that owner is not resolvable, and the child object is ignored by the garbage collector. ([#92743](https://github.com/kubernetes/kubernetes/pull/92743), [@liggitt](https://github.com/liggitt)) [SIG API Machinery, Apps and Testing]


## Changes by Kind

### Deprecation

- NFS Drivers are now enabled to collect metrics, StatFS metrics provider is used to collect the metrics. (@brahmaroutu) ([#75805](https://github.com/kubernetes/kubernetes/pull/75805), [@brahmaroutu](https://github.com/brahmaroutu))

- Fixed a regression in default configurations, which enabled `PodDisruptionConditions`
  by default, that prevented the control plane's pod garbage collector from deleting
  pods that contained duplicated field keys (environmental variables with repeated keys or
  container ports). ([#121103](https://github.com/kubernetes/kubernetes/pull/121103), [@mimowo](https://github.com/mimowo))