}, idx, *client.host)
	}
	nerrs := ng.Wait()
	if freeze {
		freezeServices()
	} else {
		unfreezeServices()
	}
	return nerrs
}

// Netperf - perform mesh style network throughput test
func (sys *NotificationSys) Netperf(ctx context.Context, duration time.Duration) []madmin.NetperfNodeResult {
	length := len(sys.allPeerClients)
	if length == 0 {
		// For single node erasure setup.

- Implemented scoring for DRA-backed extended resources. ([#134058](https://github.com/kubernetes/kubernetes/pull/134058), [@bart0sh](https://github.com/bart0sh)) [SIG Node, Scheduling and Testing]

 * writes. The {@link Cache#cleanUp} method of the returned cache will also perform maintenance, but
 * calling it should not be necessary with a high throughput cache. Only caches built with {@link
 * #removalListener removalListener}, {@link #expireAfterWrite expireAfterWrite}, {@link
 * #expireAfterAccess expireAfterAccess}, {@link #weakKeys weakKeys}, {@link #weakValues

- The scheduler implemented QueueingHint in the TaintToleration plugin, enhancing the throughput of scheduling. ([#124287](https://github.com/kubernetes/kubernetes/pull/124287), [@sanposhiho](https://github.com/sanposhiho)) [SIG Scheduling and Testing]

- DRA: scheduling pods is up to 16x faster, depending on the scenario. Scheduling throughput depends a lot on cluster utilization. It is higher for lightly loaded clusters with free resources and gets lower when the cluster utilization increases. ([#127277](https://github.com/kubernetes/kubernetes/pull/127277), [@pohly](https://github.com/pohly))...

- Scheduler cache snapshot algorithm has been optimized to improve scheduling throughput. ([#74041](https://github.com/kubernetes/kubernetes/pull/74041), [@bsalamat](https://github.com/bsalamat))

 *   <li>Lifecycle management using {@code @PostConstruct} and {@code @PreDestroy} annotations</li>
 * </ul>
 *
 * <p>Components can be registered in two ways:
 * <ul>
 *   <li>As singletons, where one instance is shared throughout the container's lifecycle</li>
 *   <li>As prototypes, where a new instance is created each time the component is requested</li>
 * </ul>
 *

- TaintNodeByCondition has been moved to Beta and is enabled by default.
- Scheduler throughput has been improved by ~50% in large clusters (>2000 nodes).

### SIG-service-catalog
- The Originating Identity feature, which lets the broker know which user that performed an action, is now GA.

* [fluentd-gcp addon] Increase CPU limit for fluentd to 1 core to achieve 100kb/s throughput. ([#62430](https://github.com/kubernetes/kubernetes/pull/62430), [@bmoyles0117](https://github.com/bmoyles0117))

 * Implementations of this interface will be discovered through the Java ServiceLoader mechanism
 * and their provided project scopes will be available throughout the Maven build process.
 * <p>
 * Example usage:
 * <pre>
 * public class CustomProjectScopeProvider implements ProjectScopeProvider {
 *     public Collection&lt;ProjectScope&gt; provides() {