* etcd3: use PrevKV to remove additional get ([#34246](https://github.com/kubernetes/kubernetes/pull/34246), [@hongchaodeng](https://github.com/hongchaodeng))
* etcd3: avoid unnecessary decoding in etcd3 client  ([#34435](https://github.com/kubernetes/kubernetes/pull/34435), [@wojtek-t](https://github.com/wojtek-t))

   * This is a big deal - we're talking factors of 2-8 in microbenchmarks, not just 10-20%. (This is
   * a hotspot optimizer bug, which should be fixed, but that's a separate, big project).
   *
   * The coding pattern above is heavily used in java.util, e.g. in ArrayList. There is a
   * RangeCheckMicroBenchmark in the JDK that was used to test this.
   *

    // two outgoing writes
    val sink = stream.sink
    sink.write(Buffer().writeUtf8("abcde"), 5)
    sink.write(Buffer().writeUtf8("fghij"), 5)
    sink.close()

    // verify the peer received one incoming frame
    assertThat(peer.takeFrame().type).isEqualTo(Http2.TYPE_HEADERS)
    val data = peer.takeFrame()
    assertThat(data.type).isEqualTo(Http2.TYPE_DATA)
    assertArrayEquals("abcdefghij".toByteArray(), data.data)

}

func getIncomingS3RequestsMD() MetricDescription {
	return MetricDescription{
		Namespace: s3MetricNamespace,
		Subsystem: requestsSubsystem,
		Name:      incomingTotal,
		Help:      "Total number of incoming S3 requests",
		Type:      gaugeMetric,
	}
}

func getS3RequestsTotalMD() MetricDescription {
	return MetricDescription{
		Namespace: s3MetricNamespace,
		Subsystem: requestsSubsystem,

- Introduce a new server run option 'shutdown-send-retry-after'. If true the HTTP Server will continue listening until all non longrunning request(s) in flight have been drained, during this window all incoming requests will be rejected with a status code `429` and a 'Retry-After' response header. ([#101257](https://github.com/kubernetes/kubernetes/pull/101257), [@tkashem](https://github.com/tkashem))

#include "tensorflow/core/graph/graph.h"
#include "tensorflow/core/graph/node_builder.h"
#include "tensorflow/core/graph/validate.h"
#include "tensorflow/core/lib/gtl/array_slice.h"
#include "tensorflow/core/platform/coding.h"
#include "tensorflow/core/platform/errors.h"
#include "tensorflow/core/platform/mem.h"
#include "tensorflow/core/platform/mutex.h"
#include "tensorflow/core/platform/protobuf.h"
#include "tensorflow/core/platform/status.h"

* Replica set conditions API ([#33905](https://github.com/kubernetes/kubernetes/pull/33905), [@kargakis](https://github.com/kargakis))
* etcd3: avoid unnecessary decoding in etcd3 client  ([#34435](https://github.com/kubernetes/kubernetes/pull/34435), [@wojtek-t](https://github.com/wojtek-t))

SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.


-------
internal/common/binary.go in the gopsutil is copied and modified from golang/encoding/binary.go.



Copyright (c) 2009 The Go Authors. All rights reserved.

Redistribution and use in source and binary forms, with or without

- Client-side apply will now use OpenAPI `v3` by default. ([#120707](https://github.com/kubernetes/kubernetes/pull/120707), [@Jefftree](https://github.com/Jefftree))
- Decoding etcd's response now respects the timeout context. ([#121614](https://github.com/kubernetes/kubernetes/pull/121614), [@HirazawaUi](https://github.com/HirazawaUi))

### Other (Cleanup or Flake)

- Kube-apiserver: sets an upper-bound on the lifetime of idle keep-alive connections and time to read the headers of incoming requests ([#103958](https://github.com/kubernetes/kubernetes/pull/103958), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Node]

## Dependencies

### Added
_Nothing has changed._