- Kube-apiserver: redirect responses are no longer returned from backends by default. Set `--aggregator-reject-forwarding-redirect=false` to continue forwarding redirect responses. ([#112193](https://github.com/kubernetes/kubernetes/pull/112193), [@jindijamie](https://github.com/jindijamie)) [SIG API Machinery and Testing]

### Other notable changes

* Fixes an issue with upgrade requests made via pod/service/node proxy subresources sending a non-absolute HTTP request-uri to backends ([#52065](https://github.com/kubernetes/kubernetes/pull/52065), [@liggitt](https://github.com/liggitt))

        isDaemon = true
        uncaughtExceptionHandler = loggingUncaughtExceptionHandler
      }
    }

  private val backend = TaskRunner.RealBackend(threadFactory)
  private val taskRunner = TaskRunner(backend)
  private val queue = taskRunner.newQueue()

  @AfterEach fun tearDown() {
    backend.shutdown()
  }

  @Test fun test() {
    val t1 = System.nanoTime() / 1e6

 */
class TaskRunner(
  val backend: Backend,
  internal val logger: Logger = TaskRunner.logger,
) : Lockable {
  private var nextQueueName = 10000
  private var coordinatorWaiting = false
  private var coordinatorWakeUpAt = 0L

  /**
   * When we need a new thread to run tasks, we call [Backend.execute]. A few microseconds later we

- Fixes an issue in winkernel proxier that causes proxy rules to leak anytime service backends are modified. ([#112840](https://github.com/kubernetes/kubernetes/pull/112840), [@daschott](https://github.com/daschott)) [SIG Network and Windows]

  if (loggingEnabled) {
    startNs = queue.taskRunner.backend.nanoTime()
    log(task, queue, "starting")
  }

  var completedNormally = false
  try {
    val result = block()
    completedNormally = true
    return result
  } finally {
    if (loggingEnabled) {
      val elapsedNs = queue.taskRunner.backend.nanoTime() - startNs
      if (completedNormally) {

    if (isLoom()) {
      val backend = TaskRunner.RealBackend(loomThreadFactory())
      val taskRunner = TaskRunner(backend)

      OkHttpClient
        .Builder()
        .connectionPool(
          buildConnectionPool(
            connectionListener = connectionListener,
            taskRunner = taskRunner,
          ),
        ).dispatcher(Dispatcher(backend.executor))
        .taskRunnerInternal(taskRunner)

 * [advanceUntil]. These functions don't return until the task threads are all idle.
 *
 * Task threads release their execution privilege in these ways:
 *
 *  * By yielding in [TaskRunner.Backend.coordinatorWait].
 *  * By yielding in [BlockingQueue.poll].
 *  * By completing.
 */
class TaskFaker : Closeable {
  val logger = Logger.getLogger("TaskFaker." + instance++)

  /**
   * Results are posted here as they occur. The find job is done when either one plan completes
   * successfully or all plans fail.
   */
  private val connectResults = taskRunner.backend.decorate(LinkedBlockingDeque<ConnectResult>())

  override fun find(): RealConnection {
    var firstException: IOException? = null
    try {
      while (tcpConnectsInFlight.isNotEmpty() || routePlanner.hasNext()) {

  The API Server will no longer proxy non-101 responses for upgrade requests. This could break proxied backends (such as an extension API server) that respond to upgrade requests with a non-101 response code. ([#92941](https://github.com/kubernetes/kubernetes/pull/92941), [@tallclair](https://github.com/tallclair)) [SIG API Machinery]