*/
    private const val MAX_QUEUE_SIZE = 16L * 1024 * 1024 // 16 MiB.

    /**
     * The maximum amount of time after the client calls [close] to wait for a graceful shutdown. If
     * the server doesn't respond the web socket will be canceled.
     */
    const val CANCEL_AFTER_CLOSE_MILLIS = 60L * 1000

    /**

        .isEqualTo("response body")
      requestBody.close()
      assertThat(responseBody.readUtf8Line()).isNull()
    }
    body.awaitSuccess()
  }

  /**
   * Duplex calls that have follow-ups are weird. By the time we know there's a follow-up we've
   * already split off another thread to stream the request body. Because we permit at most one

- Made legacy watch calls (`ResourceVersion` = 0 or unset) that generate init-events weigh higher in `API Priority and Fairness (APF)` seat usage. Properly accounting for their cost protects the API server from CPU overload. Users might see increased throttling of such calls as a result. ([#134601](https://github.com/kubernetes/kubernetes/pull/134601), [@shyamjvs](https://github.com/shyamjvs))

          if (lengthA != lengthB) return if (lengthA < lengthB) -1 else 1
          return 0
        }
      }

    /**
     * Holds interned instances. This needs to be above the init() calls below so that it's
     * initialized by the time those parts of `<clinit>()` run. Guarded by CipherSuite.class.
     */
    private val INSTANCES = mutableMapOf<String, CipherSuite>()

### Feature

- Added 3 new metrics for monitoring async API calls in the scheduler when the `SchedulerAsyncAPICalls` feature gate was enabled:
    - `scheduler_async_api_call_execution_total`: tracks executed API calls by call type and result (success/error)
    - `scheduler_async_api_call_duration_seconds`: histogram of API call execution duration by call type and result

   *
   * @since 2.0
   */
  interface Entry<E extends @Nullable Object> {

    /**
     * Returns the multiset element corresponding to this entry. Multiple calls to this method
     * always return the same instance.
     *
     * @return the element corresponding to this entry
     */
    @ParametricNullness
    E getElement();

    /**

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

        key.mid = Encdec.dec_uint16le(sbuf, 34) & 0xFFFF;

        /* 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().
         */

* Audit logs might impact the API server performance and the latency of large request and response calls. The issue is observed under the following conditions: if `AdvancedAuditing` feature gate is enabled, which is the default case, if audit logging uses the log backend in JSON format, or if the audit policy records large API calls for requests or responses.

For more information, see [#51899](https://github.com/kubernetes/kubernetes/issues/51899).

   * constructors very wisely did not accept varargs.)
   *
   * @param initialArraySize the exact size of the initial backing array for the returned array list
   *     ({@code ArrayList} documentation calls this value the "capacity")
   * @return a new, empty {@code ArrayList} which is guaranteed not to resize itself unless its size
   *     reaches {@code initialArraySize + 1}

        .isEqualTo(expectedKeys);
    assertThat(cache.asMap().keySet()).containsNoneIn(removalNotifications.keySet());
    threadPool.shutdown();
    threadPool.awaitTermination(300, SECONDS);
  }

  /**
   * Calls get() repeatedly from many different threads, and tests that all of the removed entries
   * (removed because of size limits or expiration) trigger appropriate removal notifications.
   */