- Added the metric terminated_containers_total to track the number of containers failed or succeeded broken down by exit code ([#137453](https://github.com/kubernetes/kubernetes/pull/137453), [@rawsocket](https://github.com/rawsocket)) [SIG Instrumentation, Node and Testing]

		}

		queueDirEnv := target.EnvMQTTQueueDir
		if k != config.Default {
			queueDirEnv = queueDirEnv + config.Default + k
		}

		mqttArgs := target.MQTTArgs{
			Enable:               enabled,
			Broker:               *brokerURL,
			Topic:                env.Get(topicEnv, kv.Get(target.MqttTopic)),
			QoS:                  byte(qos),
			User:                 env.Get(usernameEnv, kv.Get(target.MqttUsername)),

                    log.trace(Hexdump.toHexString(buffer, 4, n));

                }
                /*
                 * For some reason this can sometimes get broken up into another
                 * "NBSS Continuation Message" frame according to WireShark
                 */

                this.out.write(buffer, 0, 4 + n);
                this.out.flush();
            }

		writeErrorResponse(ctx, w, toAPIError(ctx, err), r.URL)
		return
	}

	etag := getDecryptedETag(formValues, objInfo, false)

	// We must not use the http.Header().Set method here because some (broken)
	// clients expect the ETag header key to be literally "ETag" - not "Etag" (case-sensitive).
	// Therefore, we have to set the ETag directly as map entry.
	w.Header()[xhttp.ETag] = []string{`"` + etag + `"`}

    server.enqueue(mockResponse)
    val responseAfter = MockResponse(body = "This comes after a busted connection")
    server.enqueue(responseAfter)
    server.enqueue(responseAfter) // Enqueue 2x because the broken connection may be reused.
    val response1 = getResponse(newRequest("/a"))
    response1.body
      .source()
      .timeout()
      .timeout(100, TimeUnit.MILLISECONDS)

* Fix `kubectl logs -f` for windows server containers. ([#81747](https://github.com/kubernetes/kubernetes/pull/81747), [@Random-Liu](https://github.com/Random-Liu))
* Fix VMSS LoadBalancer backend pools so that the network won't be broken when instances are upgraded to latest model ([#81411](https://github.com/kubernetes/kubernetes/pull/81411), [@nilo19](https://github.com/nilo19))

- HTTP/2 connection health check is enabled by default in all Kubernetes clients to fix persistently broken connections (https://github.com/kubernetes/client-go/issues/374). If needed, users can tune the feature via the HTTP2_READ_IDLE_TIMEOUT_SECONDS and HTTP2_PING_TIMEOUT_SECONDS environment variables. The feature is disabled if HTTP2_READ_IDLE_TIMEOUT_SECONDS...

- Fixed an issue in the Windows kube-proxy (winkernel) where IPv4 and IPv6 Service load balancers could be incorrectly shared, causing broken dual-stack Service behavior. The kube-proxy now tracks load balancers per IP family, enabling correct support for PreferDualStack and RequireDualStack Services on Windows nodes. ([#136373](https://github.com/kubernetes/kubernetes/pull/136373), [@...

* ✏️ Fix broken link in `docs/tutorial/sql-databases.md` in several languages. PR [#10716](https://github.com/tiangolo/fastapi/pull/10716) by [@theoohoho](https://github.com/theoohoho).
* ✏️ Remove broken links from `external_links.yml`. PR [#10943](https://github.com/tiangolo/fastapi/pull/10943) by [@Torabek](https://github.com/Torabek).

    * As with the existing Wheezy image on GCE, docker is expected to already be installed in the image.
* Mark kube-push.sh as broken ([#25095](https://github.com/kubernetes/kubernetes/pull/25095), [@ihmccreery](https://github.com/ihmccreery))