slightly more likely to do so now.
    *   Code that tries directly getting gradients with respect to symbolic
        Keras inputs/outputs. Use `GradientTape` on the actual Tensors passed to
        the already-constructed model instead.
    *   Code that requires very tricky shape manipulation via converted op
        layers in order to work, where the Keras symbolic shape inference proves

- Fix fake client to accept request without metadata.name to better emulate behavior of actual client. ([#126727](https://github.com/kubernetes/kubernetes/pull/126727), [@jpbetz](https://github.com/jpbetz)) [SIG API Machinery]

  if (forward_input) {
    TFE_DeleteTensorHandle(n);
  }

  int num_retvals = 1;
  if (async) {
    // Enqueue dummy ops so we backlog async execution & actually test async.
    // This is usually unnecessary, but we've experienced the occasional test
    // failure when testing async mode with no explicit forwarding.
    for (int i = 0; i < 100000; ++i) {

- The conformance tests:
  - Services should serve multiport endpoints from pods
  - Services should serve a basic endpoint from pods
  were only validating the API objects, not performing any validation on the actual Services implementation.

* `kubeadm join` is now blocking on the kubelet performing the TLS Bootstrap properly. Earlier, `kubeadm join` only did the discovery part and exited successfully without checking that the kubelet actually started properly and performed the TLS bootstrap correctly. Now, as kubeadm runs some post-join steps (for example, annotating the Node API object with the CRISocket), `kubeadm join` is now waiting for the kubelet to perform the TLS Bootstrap, and...

			writeErrorResponseJSON(ctx, w, toAdminAPIErr(ctx, err), r.URL)
			return
		}
		targetUser = lookupResult.NormDN
		opts.claims[ldapUser] = targetUser // username DN
		opts.claims[ldapActualUser] = lookupResult.ActualDN

		// Add LDAP attributes that were looked up into the claims.
		for attribKey, attribValue := range lookupResult.Attributes {
			opts.claims[ldapAttribPrefix+attribKey] = attribValue
		}

   * NavigableMap#subMap(Object, boolean, Object, boolean) subMap()}, {@link
   * NavigableMap#tailMap(Object, boolean) tailMap()}, and {@link NavigableMap#headMap(Object,
   * boolean) headMap()}) to actually construct the view. Consult these methods for a full
   * description of the returned view's behavior.
   *
   * <p><b>Warning:</b> {@code Range}s always represent a range of values using the values' natural

   * NavigableMap#subMap(Object, boolean, Object, boolean) subMap()}, {@link
   * NavigableMap#tailMap(Object, boolean) tailMap()}, and {@link NavigableMap#headMap(Object,
   * boolean) headMap()}) to actually construct the view. Consult these methods for a full
   * description of the returned view's behavior.
   *
   * <p><b>Warning:</b> {@code Range}s always represent a range of values using the values' natural

### Bug or Regression

- Cherry-pick #115769: Fix the problem Pod terminating stuck because of trying to umount not actual mounted dir. ([#119831](https://github.com/kubernetes/kubernetes/pull/119831), [@cartermckinnon](https://github.com/cartermckinnon)) [SIG Node and Storage]

  /**
   * The elements in this map would pass the UNDER test, but are known not to be public suffixes and
   * are thus excluded from consideration. Since it refers to elements in UNDER of the same type,
   * the type is actually not important here. The map is simply used for consistency reasons.
   */
  public static final ImmutableMap<String, PublicSuffixType> EXCLUDED =
      TrieParser.parseTrie(