- `kubectl rollout undo` now returns errors when attempting to rollback a deployment to a non-existent revision ([#70039](https://github.com/kubernetes/kubernetes/pull/70039), [@liggitt](https://github.com/liggitt))

- Fixed a bug in the scheduler where it would crash when prefilter returns a non-existent node. ([#124933](https://github.com/kubernetes/kubernetes/pull/124933), [@AxeZhan](https://github.com/AxeZhan)) [SIG Scheduling and Testing]
- Fixed a bug that Pods could stuck in the unschedulable pod pool 

// with time.  Event consumers should not rely on the timing of an event
// with a given Reason reflecting a consistent underlying trigger, or the
// continued existence of events with that Reason.  Events should be
// treated as informative, best-effort, supplemental data.
message Event {
  // Standard object's metadata.

- CloudNodeLifecycleController will check node existence status before shutdown status when monitoring nodes. ([#90737](https://github.com/kubernetes/kubernetes/pull/90737), [@jiahuif](https://github.com/jiahuif)) [SIG Apps and Cloud Provider]

    * This result into:
    * - clusters moving from IPv4 to duakstack will require no changes
    * - clusters moving from IPv6 (while not seen in the wild, we can not rule out their existence) to dualstack will require deleting backend pools (the reconciler will take care of creating correct backendpools)

hastily, just in time to avoid shrinking away altogether.

`That WAS a narrow escape!' said Alice, a good deal frightened at
the sudden change, but very glad to find herself still in
existence; `and now for the garden!' and she ran with all speed
back to the little door:  but, alas! the little door was shut
again, and the little golden key was lying on the glass table as

- Fixed bug that caused a resource to include patch directives when using strategic merge patch against a non-existent field. ([#117568](https://github.com/kubernetes/kubernetes/pull/117568), [@alexzielenski](https://github.com/alexzielenski))

* client-go: fake clients now properly return NotFound errors when attempting to patch non-existent objects ([#70886](https://github.com/kubernetes/kubernetes/pull/70886), [@bouk](https://github.com/bouk))

* Azure disks now support external resource groups.
([#64427](https://github.com/kubernetes/kubernetes/pull/64427), [@andyzhangx](https://github.com/andyzhangx))
* Added reason message logs for non-existent Azure resources.
([#64248](https://github.com/kubernetes/kubernetes/pull/64248), [@feiskyer](https://github.com/feiskyer))

### SIG CLI

* kubeadm: Fix a bug on some OSes where the kubelet tried to mount a volume path that is non-existent and on a read-only filesystem  ([#55320](https://github.com/kubernetes/kubernetes/pull/55320), [@andrewrynhard](https://github.com/andrewrynhard))