- Added a feature to report an event about a Pod if kubelet observes a failed attach operation, even if the kubelet is running with `--enable-controller-attach-detach=false`. ([#124884](https://github.com/kubernetes/kubernetes/pull/124884), [@carlory](https://github.com/carlory))

- Revert building binaries with PIE mode. ([#105352](https://github.com/kubernetes/kubernetes/pull/105352), [@ehashman](https://github.com/ehashman))
- Reverts adding namespace label to admission metrics (and histogram exansion) due to cardinality issues. ([#104033](https://github.com/kubernetes/kubernetes/pull/104033), [@s-urbaniak](https://github.com/s-urbaniak))

Bug or Regression

- Fixed a regression that could incorrectly reject pods with OutOfCpu errors if they were rapidly scheduled after other pods were reported as complete in the API. The Kubelet now waits to report the phase of a pod as terminal in the API until all running containers are guaranteed to have stopped and no new containers can be started.  Short-lived pods may take slightly longer (~1s) to report Succeeded or Failed after this change. ([#108749](https://github.com/kubern...

- Reduced GCE log rotation check from 1 hour to every 5 minutes.  Rotation policy is unchanged (new day starts, log file size > 100MB). ([#76352](https://github.com/kubernetes/kubernetes/pull/76352), [@jpbetz](https://github.com/jpbetz))

- E2e framework: added `--report-complete-ginkgo` and `--report-complete-junit` parameters. They work like `ginkgo --json-report <report dir>/ginkgo/report.json --junit-report <report dir>/ginkgo/report.xml`. ([#115678](https://github.com/kubernetes/kubernetes/pull/115678), [@pohly](https://github.com/pohly)) [SIG Testing]

* In GCE, add measures to prevent corruption of known_tokens.csv. ([#49897](https://github.com/kubernetes/kubernetes/pull/49897), [@mikedanese](https://github.com/mikedanese))

- Reverted NewVolumeManagerReconstruction and SELinuxMountReadWriteOncePod feature gates to disabled by default to resolve a regression of volume reconstruction on kubelet/node restart ([#117751](https://github.com/kubernetes/kubernetes/pull/117751), [@liggitt](https://github.com/liggitt)) [SIG Storage]

### Other notable changes

* Connections from Pods to Services with 0 endpoints will now ICMP reject immediately, rather than blackhole and timeout. ([#72534](https://github.com/kubernetes/kubernetes/pull/72534), [@thockin](https://github.com/thockin))

* The maximum number of images the Kubelet will report in the Node status can now be controlled via the Kubelet's `--node-status-max-images` flag. The default (50) remains the same. ([#64170](https://github.com/kubernetes/kubernetes/pull/64170), [@mtaufen](https://github.com/mtaufen))

- Bug fixes for EndpointSlice controller that prevent race condition and modifying shared objects. ([#87731](https://github.com/kubernetes/kubernetes/pull/87731), [@robscott](https://github.com/robscott)) [SIG Apps and Network]