- Improved scheduling performance when many nodes, and prefilter returned 1-2 nodes (e.g. daemonset)
  
  For developers of out-of-tree PostFilter plugins, note that the semantics of NodeToStatusMap are changing: A node with an absent value in the NodeToStatusMap should be interpreted as having an UnschedulableAndUnresolvable status. ([#125197](https://github.com/kubernetes/kubernetes/pull/125197), [@gabesaba](https://github.com/gabesaba))

- Fixed indexer bug that resulted in incorrect index updates if number of index values for a given object was changing during update ([#109137](https://github.com/kubernetes/kubernetes/pull/109137), [@wojtek-t](https://github.com/wojtek-t))

- Add SizeBasedListCostEstimate feature gate, enabled by default, changing method of assigning APF seats to LIST request. Assign one seat per 100KB of data loaded to memory at once to handle LIST request. ([#132932](https://github.com/kubernetes/kubernetes/pull/132932), [@serathius](https://github.com/serathius)) [SIG API...

* Parameterize session affinity timeout seconds in service API for Client IP based session affinity. ([#49850](https://github.com/kubernetes/kubernetes/pull/49850), [@m1093782566](https://github.com/m1093782566))
* Changing scheduling part of the alpha feature 'LocalStorageCapacityIsolation' to manage one single local ephemeral storage resource ([#50819](https://github.com/kubernetes/kubernetes/pull/50819), [@NickrenREN](https://github.com/NickrenREN))

- The calculation of pod UIDs for static pods has changed to ensure each static pod gets a unique value - this will cause all static pod containers to be recreated/restarted if an in-place kubelet upgrade from 1.20 to 1.21 is performed. Note that draining pods before upgrading the kubelet across minor versions is the supported upgrade path. ([#87461](https://github.com/kubernetes/kubernetes/pull/87461), [@bboreham](https://github.com/bboreham)) [SIG Node]

    that need to run prior to the network being ready.
* **Enhance Kubelet QoS**:
  * Pods are placed under a new cgroup hierarchy by default. This feature requires
    draining and restarting the node as part of upgrades. To opt-out set
    `--cgroups-per-qos=false`.
  * If `kube-reserved` and/or `system-reserved` are specified, node allocatable

* Fixed CSIDriver API object to allow missing fields. ([#69331](https://github.com/kubernetes/kubernetes/pull/69331), [@jsafrane](https://github.com/jsafrane))
* Allows changing nodeName in endpoint update. ([#68575](https://github.com/kubernetes/kubernetes/pull/68575), [@prameshj](https://github.com/prameshj))

* Update defaultbackend to 1.5: move /metrics to a private endpoint (:10254). ([#69383](https://github.com/kubernetes/kubernetes/pull/69383), [@bowei](https://github.com/bowei))
* Allows changing nodeName in endpoint update. ([#68575](https://github.com/kubernetes/kubernetes/pull/68575), [@prameshj](https://github.com/prameshj))

- Fix indexer bug that resulted in incorrect index updates if number of index values for a given object was changing during update ([#109137](https://github.com/kubernetes/kubernetes/pull/109137), [@wojtek-t](https://github.com/wojtek-t)) [SIG API Machinery]

- 'Container `preStop` and `postStart` lifecycle handlers using `httpGet` now
  honor the specified `scheme` and `headers` fields. This enables setting custom
  headers and changing the scheme to `HTTPS`, consistent with container
  startup/readiness/liveness probe capabilities. Lifecycle handlers configured
  with `scheme: HTTPS` that encounter errors indicating the endpoint is actually