- Kube-controller-manager: Fixes bug setting headless service labels on endpoints ([#85361](https://github.com/kubernetes/kubernetes/pull/85361), [@liggitt](https://github.com/liggitt))

        <iframe class="w-100 h-100 position-absolute" frameborder="0"
                src="<%=request.getContextPath()%>${serverPath}<%= response.encodeURL("/_plugin/kopf/") %>"
                seamless></iframe>
    </div>

    <jsp:include page="/WEB-INF/view/common/admin/footer.jsp"></jsp:include>
</div>
<jsp:include page="/WEB-INF/view/common/admin/foot.jsp"></jsp:include>
</body>

- fixes a validation error that could prevent updates to StatefulSet objects containing non-normalized resource requests ([#66165](https://github.com/kubernetes/kubernetes/pull/66165), [@liggitt](https://github.com/liggitt))
- Headless Services with no ports defined will now create Endpoints correctly, and appear in DNS. ([#67622](https://github.com/kubernetes/kubernetes/pull/67622), [@thockin](https://github.com/thockin))

## MinIO Deployment on Kubernetes

There are multiple options to deploy MinIO on Kubernetes:

- MinIO-Operator: Operator offers seamless way to create and update highly available distributed MinIO clusters. Refer [MinIO Operator documentation](https://github.com/minio/minio-operator/blob/master/README.md) for more details.

While containers provide isolated application execution environment, orchestration platforms allow seamless scaling by helping replicate and manage containers. MinIO extends this by adding isolated storage environment for each tenant.

    private final List<EventSpy> eventSpies;

    @Inject
    public EventSpyDispatcher(List<EventSpy> eventSpies) {
        // make copy to get rid of needless overhead for dynamic lookups
        this.eventSpies = new ArrayList<>(eventSpies);
    }

    public ExecutionListener chainListener(ExecutionListener listener) {
        if (eventSpies.isEmpty()) {

//go:build darwin || dragonfly || freebsd || linux || netbsd || openbsd || solaris

package net

import (
	"bytes"
	"syscall"
	"testing"
	"time"
)

// Test that a client can't trigger an endless loop of write system
// calls on the server by shutting down the write side on the client.
// Possibility raised in the discussion of https://golang.org/cl/71973.
func TestEndlessWrite(t *testing.T) {
	t.Parallel()

// This file defines the tf_executor dialect: it models the TensorFlow executor
// semantics and can represent arbitrary TensorFlow graphs. As such it follows
// the existing execution model that includes deadness propagation, concurrent
// semantics, and control dependencies.

#ifndef TENSORFLOW_COMPILER_MLIR_TENSORFLOW_IR_TF_EXECUTOR_H_
#define TENSORFLOW_COMPILER_MLIR_TENSORFLOW_IR_TF_EXECUTOR_H_

  // only.
  virtual void Print() const = 0;
  virtual ~DeadnessAnalysis();

  string DebugString(DeadnessPredicate predicate) const;

  // Run the deadness analysis over `graph` and returns an error or a populated
  // instance of DeadnessAnalysis in `result`.
  static Status Run(const Graph& graph,
                    std::unique_ptr<DeadnessAnalysis>* result);

 protected:

- On single-stack configured (IPv4 or IPv6, but not both) clusters, Services which are both headless (no clusterIP) and selectorless (empty or undefined selector) will report `ipFamilyPolicy RequireDualStack` and will have entries in `ipFamilies[]` for both IPv4 and IPv6.  This is a change from alpha, but does not have any impact on the manually-specified...