## Changelog since v1.22.14

## Changes by Kind

### Bug or Regression

- Kube-apiserver: gzip compression switched from level 4 to level 1 to improve large list call latencies in exchange for higher network bandwidth usage (10-50% higher). This increases the headroom before very large unpaged list calls exceed request timeout limits. ([#112401](https://github.com/kubernetes/kubernetes/pull/112401), [@shyamjvs](https://github.com/shyamjvs)) [SIG API Machinery]

- ACTION_REQUIRED

- Kube-apiserver: `gzip` compression switched from level 4 to level 1 to improve large list call latencies in exchange for higher network bandwidth usage (10-50% higher). This increases the headroom before very large unpaged list calls exceed request timeout limits. ([#112299](https://github.com/kubernetes/kubernetes/pull/112299), [@shyamjvs](https://github.com/shyamjvs))

		globalReplicationPool.Get().queueReplicaTask(roi)
	}
	return roi
}

const (
	mrfSaveInterval  = 5 * time.Minute
	mrfQueueInterval = mrfSaveInterval + time.Minute // A minute higher than save interval

	mrfRetryLimit = 3 // max number of retries before letting scanner catch up on this object version
	mrfMaxEntries = 1000000
)

func (p *ReplicationPool) persistMRF() {

- Made legacy watch calls (`ResourceVersion` = 0 or unset) that generate init-events weigh higher in `API Priority and Fairness (APF)` seat usage. Properly accounting for their cost protects the API server from CPU overload. Users might see increased throttling of such calls as a result. ([#134601](https://github.com/kubernetes/kubernetes/pull/134601),...

    *   Nominated pods are taken into account during scheduling to avoid starvation of higher priority pods. ([#55933](https://github.com/kubernetes/kubernetes/pull/55933),[ @bsalamat](https://github.com/bsalamat))

            </ul>

            <h2>Examples</h2>

            <ul>
            <li>JDK version range: {@code inrange(${java.version}, '[11,)')} (JDK 11 or higher)</li>
            <li>OS check: {@code ${os.name} == 'windows'}</li>
            <li>File existence: {@code exists('${project.basedir}/src/**}{@code /*.xsd')}</li>

      return map().floorKey(e);
    }

    @Override
    public @Nullable K ceiling(@ParametricNullness K e) {
      return map().ceilingKey(e);
    }

    @Override
    public @Nullable K higher(@ParametricNullness K e) {
      return map().higherKey(e);
    }

    @Override
    public @Nullable K pollFirst() {
      return keyOrNull(map().pollFirstEntry());
    }

    @Override

  managed by the GLBC Ingress Controller. This can cause the health checks to start failing,
  requiring you to reapply the manual edits.
  * This issue does not affect clusters that were already running Kubernetes v1.6.4 or higher.
  * This issue does not affect Health Checks that were left in the configuration
    [originally set by the GLBC Ingress Controller](https://github.com/kubernetes/ingress-nginx/tree/0.9.0-beta.1/controllers/gce#health-checks).

      control the initial parallelism setting used by autotune before the data
      pipeline has started running. The default is 16. A lower value reduces
      initial memory usage, while a higher value improves startup time.

## Keras

*  `keras.layers.experimental.DynamicEmbedding`
    * Added `DynamicEmbedding` Keras layer
    * Added 'UpdateEmbeddingCallback`