`max_io` number of concurrent requests. It is possible to adjust the `max_sleep` and `max_io` values thereby increasing the healing speed. The delays between each operation of the healer can be adjusted by the `mc admin config set alias/ heal max_sleep=1s` and maximum concurrent requests allowed before we start slowing things down can be configured with `mc admin config set alias/ heal max_io=30` . By default the wait delay is `250ms` beyond 100 concurrent operations. This means the healer will sleep...

    val elapsedNanos = System.nanoTime() - startNanos
    val elapsedMillis = TimeUnit.NANOSECONDS.toMillis(elapsedNanos)
    assertThat(elapsedMillis).isBetween(500L, 1000L)
  }

  /** Delay the response body by sleeping 1s.  */
  @Test
  fun delayResponse() {
    assumeNotWindows()
    server.enqueue(
      MockResponse
        .Builder()
        .body("ABCDEF")
        .bodyDelay(1, TimeUnit.SECONDS)
        .build(),

   *
   * <p>In order for "::" compression to work, the input should contain negative sentinel values in
   * place of the elided zeroes.
   *
   * @param hextets {@code int[]} array of eight 16-bit hextets, or -1s
   */
  private static String hextetsToIPv6String(int[] hextets) {
    // While scanning the array, handle these state transitions:
    //   start->num => "num"     start->gap => "::"

   *
   * <p>In order for "::" compression to work, the input should contain negative sentinel values in
   * place of the elided zeroes.
   *
   * @param hextets {@code int[]} array of eight 16-bit hextets, or -1s
   */
  private static String hextetsToIPv6String(int[] hextets) {
    // While scanning the array, handle these state transitions:
    //   start->num => "num"     start->gap => "::"

If you use an old version kubectl to delete a replication controller or replicaset, then after the delete command has returned, the replication controller or the replicaset will continue to exist in the key-value store for a short period of time (<1s). You probably will not notice any difference if you use kubectl manually, but you might notice it if you are using kubectl in a script.

### DELETE operation in REST API

    server.enqueue(
      MockResponse
        .Builder()
        .body("def")
        .throttleBody(1, 750, TimeUnit.MILLISECONDS)
        .build(),
    )

    // First request: time out after 1s.
    client =
      client
        .newBuilder()
        .readTimeout(Duration.ofSeconds(1))
        .build()
    executeSynchronously("/a").assertBody("abc")

    // Second request: time out after 250ms.

- Ensure Job status updates are batched by 1s. This fixes an unlikely scenario when a sequence of immediately 

read this before you upgrade)

- Added the ability to reduce both the initial delay and the maximum delay accrued between container restarts for a node for containers in `CrashLoopBackOff` across the cluster to the recommended values of `1s` initial delay and `60s` maximum delay. To set this for a node, turn on the feature gate `ReduceDefaultCrashLoopBackOffDecay`. If you are also using the feature gate `KubeletCrashLoopBackOffMax` with a configured per-node `CrashLoopBackOff.MaxContainerRestartPeriod`,...

- Fixes error handling in a kubectl method used in downstream...

- A kubelet parameter and config option has been added to change the CFS quota period from the default 100ms to some other value between 1µs and 1s. This was done to improve response latencies for workloads running in clusters with guaranteed and burstable QoS classes. ([#63437](https://github.com/kubernetes/kubernetes/pull/63437), [@szuecs](https://github.com/szuecs)) Courtesy of SIG API Machinery,...