- CSI drivers can now opt in to receive service account tokens via the secrets field instead of volume context by setting `spec.serviceAccountTokenInSecrets: true` in the CSIDriver object. This prevents tokens from being exposed in logs and other outputs. The feature is gated by the `C...

- Fixed null `lastTransitionTime` in Pod condition when setting the scheduling gate. ([#122636](https://github.com/kubernetes/kubernetes/pull/122636), [@lianghao208](https://github.com/lianghao208)) [SIG Node and Scheduling]
- Fixed recursive LIST from watch cache returning object matching key. ([#125584](https://github.com/kubernetes/kubernetes/pull/125584), [@serathius](https://github.com/serathius)) [SIG API Machinery and Testing]

* adjusted http/2 buffer sizes for apiservers to prevent starvation issues between concurrent streams ([#67902](https://github.com/kubernetes/kubernetes/pull/67902), [@liggitt](https://github.com/liggitt))
* Reduced excessive logging from fluentd-gcp-scaler. ([#68837](https://github.com/kubernetes/kubernetes/pull/68837), [@x13n](https://github.com/x13n))

* Build hyperkube in hack/local-up-cluster instead of separate binaries ([#25627](https://github.com/kubernetes/kubernetes/pull/25627), [@luxas](https://github.com/luxas))
* enable recursive processing in kubectl rollout ([#25110](https://github.com/kubernetes/kubernetes/pull/25110), [@metral](https://github.com/metral))

    /**
     * When specified as the {@code shareAccess} constructor parameter,
     * other SMB clients (including other threads making calls into jCIFS)
     * will not be permitted to access the target file and will receive "The
     * file is being accessed by another process" message.
     */
    public static final int FILE_NO_SHARE = 0x00;
    /**
     * When specified as the {@code shareAccess} constructor parameter,

* Fixes a flaw (CVE-2019-11253) in json/yaml decoding where large or malformed documents could consume excessive server resources. Request bodies for normal API requests (create/delete/update/patch operations of regular resources) are now limited to 3MB. ([#83261](https://github.com/kubernetes/kubernetes/pull/83261), [@liggitt](https://github.com/liggitt))...

	buckets := make(map[string][]MetricV2, v2MetricsMaxBuckets)
	for promMetric := range ch {
		dtoMetric := &dto.Metric{}
		err := promMetric.Write(dtoMetric)
		if err != nil {
			// Log error and continue to receive other metric
			// values
			bugLogIf(GlobalContext, err)
			continue
		}

		h := dtoMetric.GetHistogram()
		for _, b := range h.Bucket {
			labels := make(map[string]string)

- Fixed a bug where improper storage and comparison of endpoints led to excessive API traffic from the endpoints controller ([#94937](https://github.com/kubernetes/kubernetes/pull/94937), [@damemi](https://github.com/damemi)) [SIG Apps, Network and Testing]

### Other notable changes

* Fix gpu issue after node instance recreation in 1.10. ([#69595](https://github.com/kubernetes/kubernetes/pull/69595), [@jiayingz](https://github.com/jiayingz))
* Reduced excessive logging from fluentd-gcp-scaler. ([#68837](https://github.com/kubernetes/kubernetes/pull/68837), [@x13n](https://github.com/x13n))

- Reduced the number of cloud API calls and service downtime caused by excessive re-configurations of cluster LBs with externalTrafficPolicy=Local when node readiness changes (https://github.com/kubernetes/kubernetes/issues/111539). The service controller (in cloud-controller-manager) will avoid resyncing nodes which are transitioning...