}
            this.sbuf[35] = (byte) b;
        }

        /*
         * Unless key returned is null or invalid Transport.loop() always
         * calls doRecv() after and no one else but the transport thread
         * should call doRecv(). Therefore it is ok to expect that the data
         * in sbuf will be preserved for copying into BUF in doRecv().
         */

- Kube-apiserver: gzip compression switched from level 4 to level 1 to improve large list call latencies in exchange for higher network bandwidth usage (10-50% higher). This increases the headroom before very large unpaged list calls exceed request timeout limits. ([#112398](https://github.com/kubernetes/kubernetes/pull/112398), [@shyamjvs](https://github.com/shyamjvs)) [SIG API Machinery]

    protected static final String CODE = "code";

    /** Timeout for token acquisition in milliseconds. */
    protected long acquisitionTimeout = 30 * 1000L;

    /** Cache for storing group information to reduce API calls. */
    protected Cache<String, Pair<String[], String[]>> groupCache;

    /** Group cache expiry time in seconds. */
    protected long groupCacheExpiry = 10 * 60L;

## Changes by Kind

### Bug or Regression

- Kube-apiserver: gzip compression switched from level 4 to level 1 to improve large list call latencies in exchange for higher network bandwidth usage (10-50% higher). This increases the headroom before very large unpaged list calls exceed request timeout limits. ([#112401](https://github.com/kubernetes/kubernetes/pull/112401), [@shyamjvs](https://github.com/shyamjvs)) [SIG API Machinery]

* Update the GCE windows node image to include hot fixes since July. ([#81233](https://github.com/kubernetes/kubernetes/pull/81233), [@YangLu1031](https://github.com/YangLu1031))
* switch to VM Update call in attach/detach disk operation, original CreateOrUpdate call may lead to orphaned VMs or blocked resources ([#81208](https://github.com/kubernetes/kubernetes/pull/81208), [@andyzhangx](https://github.com/andyzhangx))

- Added alpha support for asynchronous Pod preemption.
  When the `SchedulerAsyncPreemption` feature gate is enabled, the scheduler now runs API calls to trigger preemptions asynchronously for better performance. ([#128170](https://github.com/kubernetes/kubernetes/pull/128170), [@sanposhiho](https://github.com/sanposhiho)) [SIG Scheduling and Testing]

* Fixes a goroutine leak in kube-apiserver when a request times out. ([#83333](https://github.com/kubernetes/kubernetes/pull/83333), [@lavalamp](https://github.com/lavalamp))
* Fix aggressive VM calls for Azure VMSS ([#83102](https://github.com/kubernetes/kubernetes/pull/83102), [@feiskyer](https://github.com/feiskyer))

		// wrote to disk - since that amounts to "copying" from a "copy"
		// instead of "copying" from source, we need the stream to be seekable
		// to ensure that we can make fan-out calls concurrently.
		buf := bytebufferpool.Get()
		defer func() {
			buf.Reset()
			bytebufferpool.Put(buf)
		}()

		md5w := md5.New()

		// Maximum allowed fan-out object size.

* Remove Job also from .status.active for Replace strategy ([#35420](https://github.com/kubernetes/kubernetes/pull/35420), [@soltysh](https://github.com/soltysh))
* Update PodAntiAffinity to ignore calls to subresources ([#35608](https://github.com/kubernetes/kubernetes/pull/35608), [@soltysh](https://github.com/soltysh))

### Bug or Regression

- Avoid GCE API calls when initializing GCE CloudProvider for Kubelets.