///

Pour éviter de bloquer la boucle d'événements, vous pouvez simplement déclarer la *fonction de chemin d'accès* avec un `def` classique au lieu de `async def`. Ainsi, FastAPI l'exécutera dans un worker de pool de threads, afin d'éviter de bloquer la boucle principale.

{* ../../docs_src/stream_data/tutorial002_py310.py ln[30:34] hl[31] *}

/// tip | Astuce

Quand vous exécutez **plusieurs processus** du même programme d'API, on les appelle couramment des **workers**.

### Processus workers et ports { #worker-processes-and-ports }

Rappelez‑vous, d'après les documents [À propos de HTTPS](https.md), qu'un seul processus peut écouter une combinaison de port et d'adresse IP sur un serveur ?

///

Um die Event-Loop nicht zu blockieren, können Sie die Pfadoperation-Funktion einfach mit normalem `def` statt `async def` deklarieren, dadurch führt FastAPI sie in einem Threadpool-Worker aus, um die Haupt-Event-Loop nicht zu blockieren.

{* ../../docs_src/stream_data/tutorial002_py310.py ln[30:34] hl[31] *}

/// tip | Tipp

- Ensure Azure availability zone is always in lower cases. ([#89722](https://github.com/kubernetes/kubernetes/pull/89722), [@feiskyer](https://github.com/feiskyer)) [SIG Cloud Provider]
- Fix failing conformance when using docker container runtime w/ the docker/journald logging driver. [#87933](https://github.com/kubernetes/kubernetes/pull/87933) ([#88586](https://github.com/kubernetes/kubernetes/pull/88586), [@jdef](https://github.com/jdef)) [SIG Node]

///

Para evitar bloquear o loop de eventos, você pode simplesmente declarar a função de operação de rota com `def` normal em vez de `async def`. Assim, o FastAPI a executará em um worker de threadpool, evitando bloquear o loop principal.

{* ../../docs_src/stream_data/tutorial002_py310.py ln[30:34] hl[31] *}

/// tip | Dica

    * Leaked File descriptors ([#275](https://github.com/docker/containerd/issues/275))
    * Additional memory overhead per container ([#21737](https://github.com/docker/docker/issues/21737))
  * Docker 1.10.3 contains [backports provided by RedHat](https://github.com/docker/docker/compare/v1.10.3...runcom:docker-1.10.3-stable) for known issues
  * Support for Docker version 1.9.x has been removed
* rkt version 1.23.0+

하지만 많은 경우, 여러 워커 프로세스를 동시에 실행하고 싶을 것입니다.

### 여러 프로세스 - 워커 { #multiple-processes-workers }

단일 프로세스가 처리할 수 있는 것보다 클라이언트가 더 많고(예: 가상 머신이 그리 크지 않을 때), 서버 CPU에 **여러 코어**가 있다면, 같은 애플리케이션을 실행하는 **여러 프로세스**를 동시에 띄우고 요청을 분산시킬 수 있습니다.

같은 API 프로그램을 **여러 프로세스**로 실행할 때, 이 프로세스들을 보통 **workers**라고 부릅니다.

### 워커 프로세스와 포트 { #worker-processes-and-ports }

* kubeadm now detects the Docker cgroup driver and starts the kubelet with the matching driver. This eliminates a common error experienced by new users in when the Docker cgroup driver is not the same as the one set for the kubelet due to different Linux distributions setting different cgroup drivers for Docker, making it hard to start the kubelet properly.

Подробнее об этом читайте в [документации Docker о shell- и exec-формах](https://docs.docker.com/reference/dockerfile/#shell-and-exec-form).

Это особенно заметно при использовании `docker compose`. См. раздел FAQ Docker Compose с техническими подробностями: [Почему мои сервисы пересоздаются или останавливаются 10 секунд?](https://docs.docker.com/compose/faq/#why-do-my-services-take-10-seconds-to-recreate-or-stop).

Докладніше про це можна прочитати в [документації Docker про shell та exec form](https://docs.docker.com/reference/dockerfile/#shell-and-exec-form).

Це може бути особливо помітно при використанні `docker compose`. Див. розділ FAQ Docker Compose для технічних деталей: [Чому мої сервіси потребують 10 секунд, щоб пересотворитися або зупинитися?](https://docs.docker.com/compose/faq/#why-do-my-services-take-10-seconds-to-recreate-or-stop).