* **Neustarts**
* Replikation (die Anzahl der laufenden Prozesse)
* **Arbeitsspeicher**
* **Schritte vor dem Start**

## Container und Docker { #containers-and-docker }

Im nächsten Kapitel über [FastAPI in Containern – Docker](docker.md){.internal-link target=_blank} werde ich einige Strategien erläutern, die Sie für den Umgang mit den anderen **Deployment-Konzepten** verwenden können.

如果您正在使用容器，例如 Docker 或 Kubernetes，我将在下一章中告诉您更多相关信息：[容器中的 FastAPI - Docker](docker.md){.internal-link target=_blank}。

比较特别的是，在 **Kubernetes** 环境中运行时，您通常**不需要**使用多个工作进程，而是**每个容器运行一个 Uvicorn 进程**。不过，我会在本章节的后续部分详细介绍这一点。

///

## 多个工作进程

您可以使用 `--workers` 命令行选项来启动多个工作进程：

//// tab | `fastapi`

如果您使用 `fastapi` 命令：

<div class="termy">

```console

})

type sharedLock struct {
	lockContext chan LockContext
}

func (ld sharedLock) backgroundRoutine(ctx context.Context, objAPI ObjectLayer, lockName string) {
	for {
		locker := objAPI.NewNSLock(minioMetaBucket, lockName)
		lkctx, err := locker.GetLock(ctx, sharedLockTimeout)
		if err != nil {
			continue
		}

	keepLock:
		for {
			select {
			case <-ctx.Done():
				return
			case <-lkctx.Context().Done():

func (restClient *ReconnectRESTClient) RLock(ctx context.Context, args LockArgs) (status bool, err error) {
	return restClient.Call("/v1/rlock", args)
}

func (restClient *ReconnectRESTClient) Lock(ctx context.Context, args LockArgs) (status bool, err error) {
	return restClient.Call("/v1/lock", args)
}

func (restClient *ReconnectRESTClient) RUnlock(ctx context.Context, args LockArgs) (status bool, err error) {

	}

	// Attempt all calls.
	_, err = lkClient.RLock(t.Context(), dsync.LockArgs{})
	if err == nil {
		t.Fatal("Expected for Rlock to fail")
	}

	_, err = lkClient.Lock(t.Context(), dsync.LockArgs{})
	if err == nil {
		t.Fatal("Expected for Lock to fail")
	}

	_, err = lkClient.RUnlock(t.Context(), dsync.LockArgs{})
	if err == nil {
		t.Fatal("Expected for RUnlock to fail")
	}

	rng := rand.New(rand.NewSource(0))
	quorum := 0
	// Numbers of unique locks
	for _, locks := range []int{100, 1000, 1e6} {
		if testing.Short() && locks > 100 {
			continue
		}
		t.Run(fmt.Sprintf("%d-locks", locks), func(t *testing.T) {
			// Number of readers per lock...
			for _, readers := range []int{1, 10, 100} {
				if locks > 1000 && readers > 1 {
					continue
				}

		return false, fmt.Errorf("Unlock attempted on a read locked entity: %s (%d read locks active)", args.Resources[0], locksHeld)
	}
	delete(l.lockMap, args.Resources[0]) // Remove the write lock
	return true, nil
}

const ReadLock = 1

func (l *lockServer) RLock(args *LockArgs) (reply bool, err error) {
	if d := atomic.LoadInt64(&l.responseDelay); d != 0 {

import org.apache.logging.log4j.core.impl.Log4jLogEvent;

/**
 * Log4j rewrite policy that converts ERROR level log events to WARN level for specified loggers.
 * This policy is useful for downgrading the severity of log events from certain loggers
 * to prevent them from being treated as critical errors.
 */
@Plugin(name = "ErrorToWarnRewritePolicy", category = Core.CATEGORY_NAME, elementType = "rewritePolicy", printObject = true)

			<DefaultRolloverStrategy fileIndex="max" min="1"
				max="${backup.max.history}" compressionLevel="9" />
		</RollingFile>
		<Rewrite name="AppFile">
			<AppenderRef ref="AppRollingFile" />
			<ErrorToWarnRewritePolicy loggers="org.apache.fontbox,org.apache.pdfbox,org.apache.poi" />
		</Rewrite>
		<RollingFile name="StatsFile" fileName="${log.file.basedir}/fess-urls.log"
			filePattern="${log.file.basedir}/fess-urls${backup.date.suffix}-%i.log.gz">

If you are using containers, for example with Docker or Kubernetes, I'll tell you more about that in the next chapter: [FastAPI in Containers - Docker](docker.md){.internal-link target=_blank}.

In particular, when running on **Kubernetes** you will probably **not** want to use workers and instead run **a single Uvicorn process per container**, but I'll tell you about it later in that chapter.

///

## Multiple Workers { #multiple-workers }