- Fixing race condition with EndpointSlice controller garbage collection. ([#91311](https://github.com/kubernetes/kubernetes/pull/91311), [@robscott](https://github.com/robscott)) [SIG Apps, Network and Testing]

- Fixed garbage collection of dangling VolumeAttachments for PersistentVolumes migrated to CSI on startup of kube-controller-manager. ([#102176](https://github.com/kubernetes/kubernetes/pull/102176), [@timebertt](https://github.com/timebertt)) [SIG Apps and Storage]

                new ElevateWord("a", 1.0f, Collections.singletonList("a"), Collections.singletonList("content"), null, null);
        ElevateWord elevateWord2 =
                new ElevateWord("b", 0.0f, Collections.singletonList("b"), Collections.singletonList("content"), null, null);
        ElevateWord elevateWord3 = new ElevateWord("c", 100.0f, Collections.singletonList("c"), Collections.singletonList("content"),

        return missingArtifacts != null && !missingArtifacts.isEmpty();
    }

    public List<Artifact> getMissingArtifacts() {
        return missingArtifacts == null ? Collections.emptyList() : Collections.unmodifiableList(missingArtifacts);
    }

    public MetadataResolutionResult addMissingArtifact(Artifact artifact) {
        missingArtifacts = initList(missingArtifacts);

        return missingArtifacts != null && !missingArtifacts.isEmpty();
    }

    public List<Artifact> getMissingArtifacts() {
        return missingArtifacts == null ? Collections.emptyList() : Collections.unmodifiableList(missingArtifacts);
    }

    public ArtifactResolutionResult addMissingArtifact(Artifact artifact) {
        missingArtifacts = initList(missingArtifacts);

### API Change

- Kubernetes 1.23 is now built with go1.19.4. To match behavior of previous Kubernetes 1.23 patch releases:
  - `kube-apiserver` defaults the GOGC setting to 63, to approximate go1.17 garbage collection memory performance in heavily loaded API servers

 *
 * <p>For example, to test {@link java.util.Collections#unmodifiableList(java.util.List)
 * Collections.unmodifiableList}'s iterator:
 *
 * {@snippet :
 * List<String> expectedElements =
 *     Arrays.asList("a", "b", "c", "d", "e");
 * List<String> actualElements =
 *     Collections.unmodifiableList(
 *         Arrays.asList("a", "b", "c", "d", "e"));

users can delete their corresponding node object by patching themselves with an OwnerReference to a cluster-scoped resource. If the OwnerReference resource does not exist or is subsequently deleted, the given node object will be deleted via garbage collection. By default, node users are authorized for create and patch requests but not delete requests against their node object. Since the NodeRestriction admission controller does not prevent patching OwnerReferences, a compromised node could leverage this...

  1. Populate new `RuntimeHandler` field in CRI's `ImageSpec` struct during image pulls from container runtimes.
  2. Pass `runtimeHandler` field in `RemoveImage()` call to container runtime in `kubelet`'s image garbage collection. ([#121456](https://github.com/kubernetes/kubernetes/pull/121456), [@kiashok](https://github.com/kiashok))

        // Full NTLM handshake testing would require more complex mocking

        // Arrange - Mock a server that supports NTLM
        mockResponse(HTTP_UNAUTHORIZED, "Unauthorized", Collections.singletonMap("WWW-Authenticate", Collections.singletonList("NTLM")),
                new ByteArrayInputStream(new byte[0]));

        // Act - Trigger handshake
        int responseCode = ntlmConnection.getResponseCode();