in early_s in Estimator
    *   Canned Estimators will now use keras optimizers by default. An error
        will be raised if tf.train.Optimizers are used, and you will have to
        switch to tf.keras.optimizers or tf.compat.v1 canned Estimators.
    *   A checkpoint converter for canned Estimators has been provided to
        transition canned Estimators that are warm started from

* kubeadm has removed `.Etcd.SelfHosting`...

    if (value != null) {
      return value;
    }

    // Store a local reference to the cache entry. If the backing map is immutable, this,
    // in combination with immutable cache entries, will ensure a thread-safe cache.
    CacheEntry<K, V> entry;

    // Check cache. We use == on purpose because it's cheaper and a cache miss is ok.
    entry = cacheEntry1;
    if (entry != null && entry.key == key) {
      return entry.value;

### SIG Scheduling

- The equivalence class cache has been redesigned to be a two level cache, resulting in a significant increase in scheduling throughput and performance. ([#65714](https://github.com/kubernetes/kubernetes/pull/65714), [@resouer](https://github.com/resouer))

    server!

 *  **Cache improvements.** This release fixes some severe cache problems
    including a bug where the cache could be corrupted upon certain access
    patterns. We also fixed a bug where the cache was being cleared due to a
    corrupted journal. We've added APIs to configure a request's `Cache-Control`
    headers, and to manually clear the cache.

- Bumped `ProcMountType` feature to on by default beta ([#130798](https://github.com/kubernetes/kubernetes/pull/130798), [@haircommander](https://github.com/haircommander)) [SIG Node]
- Calculated pod resources are now cached when adding pods to NodeInfo in the scheduler framework, improving performance when processing unschedulable pods. ([#129635](https://github.com/kubernetes/kubernetes/pull/129635), [@macsko](https://github.com/macsko)) [SIG Scheduling]

        run: |
          mkdir -p ~/.mimir
          cp .github/ci-extensions.xml ~/.m2/extensions.xml
          cp .github/ci-mimir-daemon.properties ~/.mimir/daemon.properties

      - name: Handle Mimir caches
        uses: actions/cache@0400d5f644dc74513175e3cd8d07132dd4860809 # v4
        with:
          path: ~/.mimir/local
          key: mimir-${{ runner.os }}-initial-${{ hashFiles('**/pom.xml') }}
          restore-keys: |

import java.util.concurrent.Callable
import java.util.concurrent.Executors
import kotlin.test.assertFailsWith
import okhttp3.TestUtil.threadFactory
import okhttp3.internal.cache2.Relay.Companion.edit
import okhttp3.internal.cache2.Relay.Companion.read
import okio.Buffer
import okio.ByteString
import okio.ByteString.Companion.encodeUtf8
import okio.Pipe
import okio.Source
import okio.buffer
import okio.source

* kubelet now accepts a --cni-cache-dir option, which defaults to /var/lib/cni/cache, where CNI stores cache files. ([#78908](https://github.com/kubernetes/kubernetes/pull/78908), [@dcbw](https://github.com/dcbw))

import java.util.Map;
import java.util.Map.Entry;
import java.util.Set;
import org.jspecify.annotations.Nullable;

/**
 * A map-like data structure that wraps a backing map and caches values while iterating through
 * {@link #unmodifiableKeySet()}. By design, the cache is cleared when this structure is mutated. If
 * this structure is never mutated, it provides a thread-safe view of the backing map.
 *