* Fix memory cgroup notifications, and reduce associated log spam. ([#63220](https://github.com/kubernetes/kubernetes/pull/63220), [@dashpole](https://github.com/dashpole))

the `"mixed_float16"` policy and no loss scale for other policies. In `Model.compile`, if the model's policy had a loss scale, the optimizer would be wrapped with a `LossScaleOptimizer`. With the non-experimental `Policy`, there is no loss scale associated with the `Policy`, and `Model.compile` wraps the optimizer with a `LossScaleOptimizer` if and only if the policy is a `"mixed_float16"` policy. If you previously passed a `LossScale` to the experimental `Policy`, consider just removing it, as the...

  This is achieved by reducing the TIME-WAIT state of the connection to 1 second, instead of the defaults 60 seconds. This allows kubelet to free the socket, and free conntrack entry and ephemeral port associated. ([#115143](https://github.com/kubernetes/kubernetes/pull/115143), [@aojea](https://github.com/aojea)) [SIG Network and Node]

        val uuidToTestCoverage = model.stages.flatMap { it.functionalTests }.associateBy { it.uuid }
        val testCoverageAndBuckets: List<Map<String, Any>> = objectMapper.readValue(testBucketsJson.readText())
        testCoverageAndBuckets.associate { testCoverageAndBucket ->

    fun Project.from(sourceSet: SourceSet) {
        associate(tasks.named<JavaCompile>(sourceSet.getCompileTaskName("java")))
        plugins.withType<GroovyBasePlugin>() {
            associate(tasks.named<GroovyCompile>(sourceSet.getCompileTaskName("groovy")))
        }
        plugins.withId("org.jetbrains.kotlin.jvm") {
            associate(tasks.named<KotlinCompile>(sourceSet.getCompileTaskName("kotlin")))
        }
    }

                parameters.testPathToBinaryResultsDirs = allTasks.filterIsInstance<Test>().associate { it.path to it.binaryResultsDirectory.get().asFile }

                val taskPathToReports = ******@****.***ks
                    .associate { it.path to it.genericHtmlReports() + it.findTraceJson() }
                    .filter { it.value.isNotEmpty() }

- Add a new `vSphere` metric: `cloudprovider_vsphere_vcenter_versions`. Its content shows `vCenter` hostnames with the associated server version. ([#94526](https://github.com/kubernetes/kubernetes/pull/94526), [@Danil-Grigorev](https://github.com/Danil-Grigorev)) [SIG Cloud Provider and Instrumentation]

"repelling" pods from nodes by default, which enables use cases like dedicated nodes and reserving nodes with special features for pods that need those features. We've also added a `NoExecute` taint type that evicts already-running pods, and an associated `tolerationSeconds` field to tolerations to delay the eviction for a specified amount of time. As before, taints are created using `kubectl taint` (but internally they are now represented as a field `taints` in the NodeSpec rather than using the...

to the autoscalers in other clusters if required.

### Node Components

#### Autoscaling and Metrics

* Support for custom metrics in the Horizontal Pod Autoscaler is now at v1beta1. The associated metrics APIs (custom metrics and resource/master metrics) were also moved to v1beta1. For more information, see [Before Upgrading](#before-upgrading).

* `metrics-server` is now the recommended way to provide the resource

- Fixed the `observedGeneration` field in pod resize conditions to accurately reflect the associated pod generation when both `InPlacePodVerticalScaling` and `PodObservedGenerationTracking` feature gates are enabled. ([#131157](https://github.com/kubernetes/kubernetes/pull/131157), [@natasha41575](https://github.com/natasha41575))