Dennoch besteht in beiden Fällen eine gute Chance, dass **FastAPI** [immer noch schneller](index.md#performanz){.internal-link target=_blank} als Ihr bisheriges Framework (oder zumindest damit vergleichbar) ist.

### Abhängigkeiten

* Support metrics API in `kubectl top` commands. ([#56206](https://github.com/kubernetes/kubernetes/pull/56206), [@brancz](https://github.com/brancz))

* Bump GCE metadata proxy to v0.1.9 to pick up security fixes. ([#58221](https://github.com/kubernetes/kubernetes/pull/58221), [@ihmccreery](https://github.com/ihmccreery))

- Added `kubelet_container_log_size_bytes` metric representing the log file size of a container. ([#70749](https://github.com/kubernetes/kubernetes/pull/70749), [@brancz](https://github.com/brancz))

		}

		versioned := vcfg != nil && vcfg.Versioned(item.objectPath())
		objInfos := make([]ObjectInfo, len(fivs.Versions))
		for i, fi := range fivs.Versions {
			objInfos[i] = fi.ToObjectInfo(item.bucket, item.objectPath(), versioned)
		}
		sizeS := sizeSummary{}
		for _, tier := range globalTierConfigMgr.ListTiers() {
			if sizeS.tiers == nil {
				sizeS.tiers = make(map[string]tierStats)
			}

			panic("bytes: Join output length overflow")
		}
		n += len(sep) * (len(s) - 1)
	}
	for _, v := range s {
		if len(v) > maxInt-n {
			panic("bytes: Join output length overflow")
		}
		n += len(v)
	}

	b := bytealg.MakeNoZero(n)[:n:n]
	bp := copy(b, s[0])
	for _, v := range s[1:] {
		bp += copy(b[bp:], sep)
		bp += copy(b[bp:], v)
	}
	return b
}

	// the real address and updates the immediate, using a trampoline in
	// the case where the address is not directly reachable.
	CALL	asmtest(SB)				// ef000000
	JMP	asmtest(SB)				// 6f000000

	// Branch pseudo-instructions
	BEQZ	X5, 2(PC)				// 63840200
	BGEZ	X5, 2(PC)				// 63d40200
	BGT	X5, X6, 2(PC)				// 63445300
	BGTU	X5, X6, 2(PC)				// 63645300
	BGTZ	X5, 2(PC)				// 63445000

- Fixed a race condition on API server startup ensuring previously created webhook configurations are effective before the first write request is admitted. ([#95783](https://github.com/kubernetes/kubernetes/pull/95783), [@roycaihw](https://github.com/roycaihw)) [SIG API Machinery]

- Added a new `tolerance` field to HorizontalPodAutoscaler, overriding the cluster-wide default. Enabled via the HPAConfigurableTolerance alpha feature gate. ([#130797](https://github.com/kubernetes/kubernetes/pull/130797), [@jm-franc](https://github.com/jm-franc)) [SIG API Machinery, Apps, Autoscaling, Etcd, Node, Scheduling and Testing]

		return nil, nil, err
	}

	// Wait up to 10 seconds for disks to come online.
	pools := obj.(*erasureServerPools)
	t := time.Now()
	for _, pool := range pools.serverPools {
		for _, sets := range pool.erasureDisks {
			for _, s := range sets {
				if !s.IsLocal() {
					for !s.IsOnline() {
						time.Sleep(100 * time.Millisecond)
						if time.Since(t) > 10*time.Second {

* Fix data race in node lifecycle controller ([#60831](https://github.com/kubernetes/kubernetes/pull/60831), [@resouer](https://github.com/resouer))
* Nodes are not deleted from kubernetes anymore if node is shutdown in Openstack. ([#59931](https://github.com/kubernetes/kubernetes/pull/59931), [@zetaab](https://github.com/zetaab))