Guillaume Nodet <******@****.***> 1729859506 +0200

  - Clone `kubernetes/kubernetes` in `update-vendor.sh` shallowly, instead of fetching all revisions
  - Speed up binpacking by reducing the number of PreFilter calls (call once per pod instead of #pods*#nodes times)
  - Speed up finding unneeded nodes by 5x+ in very large clusters by reducing the number of PreFilter calls

- CRI-API: IPs returned by `PodSandboxNetworkStatus`` are ignored by the kubelet for host-network pods. ([#106715](https://github.com/kubernetes/kubernetes/pull/106715), [@aojea](https://github.com/aojea))
- Call `NodeExpand` on all nodes in case of RWX volumes ([#108693](https://github.com/kubernetes/kubernetes/pull/108693), [@gnufied](https://github.com/gnufied))

    runner.admin().indices().prepareDelete("_all").execute().actionGet();
}

@AfterClass
public static void afterClass() {
    runner.close();
    runner.clean();
}
```

### Testing Best Practices

- Minimize `Thread.sleep()` - use only when absolutely necessary
- For timestamp tests: use minimal delays (50-100ms)
- Clean only test-specific indices (not `_all` unless needed)

## Changelog since v1.18.0-alpha.2

### Deprecation

- Remove all the generators from kubectl run. It will now only create pods. Additionally, deprecates all the flags that are not relevant anymore. ([#87077](https://github.com/kubernetes/kubernetes/pull/87077), [@soltysh](https://github.com/soltysh)) [SIG Architecture, SIG CLI, and SIG Testing]

returned that allows the client to continue the previous list call repeatedly until all results are retrieved.  The resulting list is identical to a list call that does not perform chunking thanks to capabilities provided by etcd3.  This allows the server to use less memory and CPU responding with very large lists.  This feature is gated as APIListChunking and is not enabled by default.  The 1.9 release will begin using this by default from all informers. ([#48921](https://github.com/kubernetes/kubernetes/pull/48921),...

        }
      }
    } else {
      /*
       * We'll call the user callback or collect the values only when all inputs complete,
       * regardless of whether some failed. This lets us avoid calling expensive methods like
       * Future.get() when we don't need to (specifically, for whenAllComplete().call*()), and it
       * lets all futures share the same listener.
       *

    void testHasMoreElementsBecomeFalse() {
        // Set small amounts so everything fits in one message
        transaction.setParameterBytesWritten(10);
        transaction.setDataBytesWritten(10);

        // First call processes all data
        transaction.nextElement();

        // Should have no more elements since all data fit in first message

      if (proxiesOrNull.isNullOrEmpty()) return immutableListOf(Proxy.NO_PROXY)

      return proxiesOrNull.toImmutableList()
    }

    call.eventListener.proxySelectStart(call, url)
    proxies = selectProxies()
    nextProxyIndex = 0
    call.eventListener.proxySelectEnd(call, url, proxies)
  }

  /** Returns true if there's another proxy to try. */
  private fun hasNextProxy(): Boolean = nextProxyIndex < proxies.size

	all = make([]*peerRESTClient, len(hosts))
	for i, host := range hosts {
		if host == nil {
			continue
		}
		all[i] = newPeerRESTClient(host, endpoints.FindGridHostsFromPeer(host))
		remote = append(remote, all[i])
	}
	if len(all) != len(remote)+1 {
		peersLogIf(context.Background(), fmt.Errorf("Expected number of all hosts (%v) to be remote +1 (%v)", len(all), len(remote)), logger.WarningKind)
	}