- Fixes an issue that can result in namespaced custom resources being orphaned when their namespace is deleted, if the CRD defining the custom resource is removed concurrently with namespaces being deleted, then recreated. ([#93790](https://github.com/kubernetes/kubernetes/pull/93790), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps]

    // We don't really care which values the exception message contains, but they should be
    // different from each other. If buildKeepingLast() collapsed duplicates, that might end up not
    // being true.
    Pattern pattern = Pattern.compile("Multiple entries with same key: four=(.*) and four=(.*)");
    assertThat(expected).hasMessageThat().matches(pattern);
      Matcher matcher = pattern.matcher(expected.getMessage());

	InitCapacity  uint64 `json:"initCapacity" msg:"ic"`   // Pool capacity at the start of rebalance

	Buckets           []string      `json:"buckets" msg:"bus"`           // buckets being rebalanced or to be rebalanced
	RebalancedBuckets []string      `json:"rebalancedBuckets" msg:"rbs"` // buckets rebalanced
	Bucket            string        `json:"bucket" msg:"bu"`             // Last rebalanced bucket

  // for updates. During a rolling update, all pods from ordinal Replicas-1 to
  // Partition are updated. All pods from ordinal Partition-1 to 0 remain untouched.
  // This is helpful in being able to do a canary based deployment. The default value is 0.
  // +optional
  optional int32 partition = 1;

  // The maximum number of pods that can be unavailable during the update.

* Add a label which prevents a node from being added to a cloud load balancer ([#53146](https://github.com/kubernetes/kubernetes/pull/53146), [@brendandburns](https://github.com/brendandburns))

container processes if an out-of-memory event occurs. This implementation has shortcomings: for `Guaranteed` pods, memory can not be fully reserved, and the page cache is at risk of being recycled. For `Burstable` pods, overcommitting memory (setting `request` less than `limit` ) could increase the risk of a container being killed when the Linux kernel detects an out of memory condition.

As an alpha feature, Kubernetes v1.22 can use the cgroups v2 API to control memory allocation and isolation....

	DeploymentID string
	Replicated   bool // true if already participating in site replication
	Empty        bool // true if cluster has no buckets
}

// getSiteStatuses gathers more info on the sites being added
func (c *SiteReplicationSys) getSiteStatuses(ctx context.Context, sites ...madmin.PeerSite) (psi []PeerSiteInfo, err error) {
	psi = make([]PeerSiteInfo, 0, len(sites))
	for _, v := range sites {

- **Network**
  - [stable] Services can reference another service by DNS name, rather than being hosted in pods ([kubernetes/features#33](https://github.com/kubernetes/enhancements/issues/33))

* The AWS credential provider can now obtain ECR credentials even without the AWS cloud provider or being on an EC2 instance. Additionally, AWS credential provider caching has been improved to honor the ECR credential timeout. ([#75587](https://github.com/kubernetes/kubernetes/pull/75587), [@tiffanyfay](https://github.com/tiffanyfay))

    match_any = [
        ":macos_x86_64_with_framework_shared_object",
        ":macos_arm64_with_framework_shared_object",
    ],
)

# Config setting that is satisfied when TensorFlow is being built with CUDA
# support through e.g. `--config=cuda` (or `--config=cuda_clang` in OSS).
alias(
    name = "is_cuda_enabled",
    actual = if_oss(
        "@local_config_cuda//:is_cuda_enabled",