- kube-proxy: a configuration file specified via `--config` is now loaded with strict deserialization, which fails if the config file contains duplicate or unknown fields. This protects against accidentally running with config files that are malformed, mis-indented, or have typos in field names, and getting unexpected behavior....

- A PersistentVolumeClaim may not have been synced to the controller local cache in time if the PersistentVolumeis bound by an external PV binder (such as kube-scheduler), so Kubernetes will now double check if PVC is not found in order to prevent the volume from being incorrectly reclaimed. ([#67062](https://github.com/kubernetes/kubernetes/pull/67062), [@cofyc](https://github.com/cofyc))...

- Content-type and verb for request metrics are now bounded to a known set. ([#89451](https://github.com/kubernetes/kubernetes/pull/89451), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery and Instrumentation]

- Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins).
  The incorrect loop logic might lead to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which is unexpected. ([#122368](https://github.com/kubernetes/kubernetes/pull/122368), [@caohe](https://github.com/caohe)) [SIG Scheduling]

- Use network.Interface.VirtualMachine.ID to get the binded VM
  Skip standalone VM when reconciling LoadBalancer ([#97635](https://github.com/kubernetes/kubernetes/pull/97635), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider]

* Audit logging with buffering enabled can increase apiserver memory usage (e.g. up to 200MB in 100-node cluster). The increase is bounded by the buffer size (configurable). Ref: issue [#60500](https://github.com/kubernetes/kubernetes/pull/60500) ([#61118](https://github.com/kubernetes/kubernetes/pull/61118), [@shyamjvs](https://github.com/shyamjvs))

    * [event-exporter addon] Bump event-exporter to 0.2.3 to pick up security fixes.
* PVC may not be synced to controller local cache in time if PV is bound by external PV binder (e.g. kube-scheduler), double check if PVC is not found to prevent reclaiming PV wrongly. ([#67062](https://github.com/kubernetes/kubernetes/pull/67062), [@cofyc](https://github.com/cofyc))

* Performance test framework and basic tests for the IPAM controller, to simulate behavior of the four supported modes under lightly loaded and loaded conditions, where load is defined as the number of operations to perform as against the configured kubernetes. ([#61143](https://github.com/kubernetes/kubernetes/pull/61143), [@satyasm](https://github.com/satyasm))

* Fix panic in kubectl cp command ([#75037](https://github.com/kubernetes/kubernetes/pull/75037), [@soltysh](https://github.com/soltysh))
* iscsi modules haven't even been loaded /sys/class/iscsi_host directory won't exist ([#74787](https://github.com/kubernetes/kubernetes/pull/74787), [@jianglingxia](https://github.com/jianglingxia))

- Fixed a regression since 1.24 in the scheduling framework when overriding MultiPoint plugins (e.g. default plugins).
  The incorrect loop logic might lead to a plugin being loaded multiple times, consequently preventing any Pod from being scheduled, which is unexpected. ([#122366](https://github.com/kubernetes/kubernetes/pull/122366), [@caohe](https://github.com/caohe)) [SIG Scheduling]