}

  @VisibleForTesting
  @WeakOuter
  final class ValueSet extends Sets.ImprovedAbstractSet<V> {
    /*
     * We currently use a fixed load factor of 1.0, a bit higher than normal to reduce memory
     * consumption.
     */

    @ParametricNullness private final K key;
    @VisibleForTesting @Nullable ValueEntry<K, V>[] hashTable;
    private int size = 0;
    private int modCount = 0;

      //
      // We use two volatile reads. We could reduce this to one by
      // putting our fields into a holder class, but (at least on x86)
      // the extra memory consumption and indirection are more
      // expensive than the extra volatile reads.
      long nanos = expirationNanos;
      long now = System.nanoTime();
      if (nanos == 0 || now - nanos >= 0) {

 * guarantees.
 *
 * <p>This class should not be assumed to be universally superior to {@code java.util.HashMap}.
 * Generally speaking, this class reduces object allocation and memory consumption at the price of
 * moderately increased constant factors of CPU. Only use this class when there is a specific reason
 * to prioritize memory over CPU.
 *
 * @author Louis Wasserman
 * @author Jon Noack
 */

  /*
   * The maximum number of bits in a square root for which we'll precompute an explicit half power
   * of two. This can be any value, but higher values incur more class load time and linearly
   * increasing memory consumption.
   */
  @VisibleForTesting static final int SQRT2_PRECOMPUTE_THRESHOLD = 256;

  @VisibleForTesting
  static final BigInteger SQRT2_PRECOMPUTED_BITS =

   * list (counter-intuitive, but mathematically consistent).
   *
   * <p><i>Performance notes:</i> while the cartesian product of sets of size {@code m, n, p} is a
   * set of size {@code m x n x p}, its actual memory consumption is much smaller. When the
   * cartesian set is constructed, the input sets are merely copied. Only as the resulting set is
   * iterated are the individual lists created, and these are not retained after iteration.
   *

environment variables into a container’s command-line args.
  * Stable version of Heapster which scales up to 1000 nodes: more metrics, reduced
latency, reduced cpu/memory consumption (~4mb per monitored node).
  * Pods now have a security context which allows users to specify:
     * attributes which apply to the whole pod:
        * User ID
        * Whether all containers should be non-root

* fix leaking memory backed volumes of terminated pods ([#36779](https://github.com/kubernetes/kubernetes/pull/36779), [@sjenning](https://github.com/sjenning))
* Default logging subsystem's resiliency was greatly improved, fluentd memory consumption and OOM error probability was reduced. ([#37021](https://github.com/kubernetes/kubernetes/pull/37021), [@Crassirostris](https://github.com/Crassirostris))



# v1.5.0-beta.1

kubelet now allows the use of XFS quotas (on XFS and suitably configured ext4fs filesystems) to monitor storage consumption for ephemeral storage.  This method of monitoring consumption, which is currently available only for `emptyDir` volumes, is faster and more accurate than the old method of walking the filesystem tree. Note that it does not enforce limits, it only monitors consumption. To utilize this functionality, set the feature gate `LocalStorageCapacityIsolationFSQuotaMonitoring=true`. For...

* Removed metrics api group ([#26073](https://github.com/kubernetes/kubernetes/pull/26073), [@piosz](https://github.com/piosz))
* Fixed check in kubectl autoscale: cpu consumption can be higher than 100%. ([#26162](https://github.com/kubernetes/kubernetes/pull/26162), [@jszczepkowski](https://github.com/jszczepkowski))

- Reduced CPU and memory consumption of `kube-apiserver` if OpenAPI V2 will not be accessed by any client. Also improved performance of the apiserver on installation of many CRDs. ([#118212](https://github.com/kubernetes/kubernetes/pull/118212), [@Jefftree](https://github.com/Jefftree))...