HTTP/2 to 16 MiB of unacknowledged data per connection. With this fix there is a limit of 16 MiB
    of unacknowledged data per stream and no per-connection limit.

 *  Fix: Don't operate on a connection after it's been returned to the pool. This race occurred
    on failed web socket connection attempts.

 *  Upgrade: [Okio 3.6.0][okio_3_6_0].

 *  Upgrade: [Kotlin 1.8.21][kotlin_1_8_21].

  /**
   * Guides the allowed concurrency among update operations. Used as a hint for internal sizing. The
   * table is internally partitioned to try to permit the indicated number of concurrent updates
   * without contention. Because assignment of entries to these partitions is not necessarily
   * uniform, the actual concurrency observed may vary. Ideally, you should choose a value to

* Make addon-manager cross-platform and use it with hyperkube ([#25631](https://github.com/kubernetes/kubernetes/pull/25631), [@luxas](https://github.com/luxas))
* kubelet: Optionally, have kubelet exit if lock file contention is observed, using --exit-on-lock-contention flag ([#25596](https://github.com/kubernetes/kubernetes/pull/25596), [@derekparker](https://github.com/derekparker))

   * of reads may not be immediately reflected on the algorithm's data structures. These structures
   * are guarded by a lock and operations are applied in batches to avoid lock contention. The
   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *

   * of reads may not be immediately reflected on the algorithm's data structures. These structures
   * are guarded by a lock and operations are applied in batches to avoid lock contention. The
   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *

   * of reads may not be immediately reflected on the algorithm's data structures. These structures
   * are guarded by a lock and operations are applied in batches to avoid lock contention. The
   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *

func gridLogOnceIf(ctx context.Context, err error, id string, errKind ...interface{}) {
	logger.LogOnceIf(ctx, "grid", err, id, errKind...)
}

// A Connection is a remote connection.
// There is no distinction externally whether the connection was initiated from
// this server or from the remote.
type Connection struct {
	// NextID is the next ID that can be used (atomic).
	NextID uint64

	// LastPong is last pong time (atomic)

 */
interface Connection {
  /** Returns the route used by this connection. */
  fun route(): Route

  /**
   * Returns the socket that this connection is using. Returns an
   * [SSL socket][javax.net.ssl.SSLSocket] if this connection is HTTPS. If this is an HTTP/2
   * connection the socket may be shared by multiple concurrent calls.
   */
  fun socket(): Socket

  // More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata
  // +optional
  optional k8s.io.apimachinery.pkg.apis.meta.v1.ObjectMeta metadata = 1;

  // Spec defines the limits enforced.
  // More info: https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
  // +optional
  optional LimitRangeSpec spec = 2;
}

* Add a flag allowing contention profiling of the API server ([#36756](https://github.com/kubernetes/kubernetes/pull/36756), [@gmarek](https://github.com/gmarek))