* ReplicationController now shares its underlying controller implementation with ReplicaSet to reduce the maintenance burden going forward. However, they are still separate resources and there should be no externally visible effects from this change. ([#49429](https://github.com/kubernetes/kubernetes/pull/49429), [@enisoc](https://github.com/enisoc))

- The `apiserver_request_total` metric is promoted to stable status and no longer has a content-type dimensions, so any alerts/charts which presume the existence of this will fail. This is however, unlikely to be the case since it was effectively an unbounded dimension in the first place. ([#99788](https://github.com/kubernetes/kubernetes/pull/99788), [@logicalhan](https://github.com/logicalhan))

* Ensured IP is set for Azure internal load balancer. ([#59083](https://github.com/kubernetes/kubernetes/pull/59083), [@feiskyer](https://github.com/feiskyer))

              @Override
              public Integer call() throws Exception {
                enterLatch.countDown();
                try {
                  new CountDownLatch(1).await(); // wait forever
                  throw new AssertionError();
                } catch (InterruptedException e) {
                  interruptedExceptionThrown.set(true);
                  throw e;
                } finally {

- Failed pods associated with a job with `parallelism = 1` are recreated by the job controller honoring exponential backoff delay again. However, for jobs with `parallelism > 1`, pods might be created without exponential backoff delay. ([#115027](https://github.com/kubernetes/kubernetes/pull/115027), [@nikhita](https://github.com/nikhita)) [SIG Apps]

Pull Requests (PRs) drive quality in MinIO’s codebase by:
- Enabling peer review without pair programming.
- Documenting changes for future reference.
- Ensuring commits tell a clear story of development.

**A poor commit lasts forever, even if code is refactored.**

## Crafting a Quality PR

A strong MinIO PR:
- Delivers a complete, valuable change (feature, bug fix, or improvement).

 * <p>
 * The cache behavior is determined by the cache retention specified in the request's metadata.
 * Results can be cached at different policies (forever, session, request, or not at all)
 * based on the {@link CacheRetention} associated with the request.
 *
 * @since 4.0.0
 * @see CacheMetadata
 * @see RequestCacheFactory
 */
@Experimental

  - `awsElasticBlockStore` for `ebs.csi.aws.com`
  - `azureDisk` for `disk.csi.azure.com`
  - `gcePersistentDisk` for `pd.csi.storage.googleapis.com`
  - `cinder` for `cinder.csi.openstack.org`
  - `csi`
  However it was still enforced using a limit in CSINode objects. ([#126924](https://github.com/kubernetes/kubernetes/pull/126924), [@carlory](https://github.com/carlory))

      taskRunner.notify()
    }

    /**
     * Wait a duration in nanoseconds. Unlike [java.lang.Object.wait] this interprets 0 as
     * "don't wait" instead of "wait forever".
     */
    @Throws(InterruptedException::class)
    @Suppress("PLATFORM_CLASS_MAPPED_TO_KOTLIN")
    override fun coordinatorWait(
      taskRunner: TaskRunner,
      nanos: Long,
    ) {

				if unlock, ok := val.(chan struct{}); ok && unlock != nil {
					// Wait until unfrozen.
					select {
					case <-unlock:
					case <-r.Context().Done():
						// if client canceled we don't need to wait here forever.
						return
					}
				}
			}
		}

		globalHTTPStats.addRequestsInQueue(1)
		pool := globalAPIConfig.getRequestsPool()
		if pool == nil {
			globalHTTPStats.addRequestsInQueue(-1)