리눅스 컨테이너를 사용하는 데에는 **보안**, **반복 가능성**, **단순함** 등의 장점이 있습니다.

/// tip | "팁"

시간에 쫓기고 있고 이미 이런것들을 알고 있다면 [`Dockerfile`👇](#build-a-docker-image-for-fastapi)로 점프할 수 있습니다.

///

<details>
<summary>도커파일 미리보기 👀</summary>

```Dockerfile
FROM python:3.9

WORKDIR /code

COPY ./requirements.txt /code/requirements.txt

* specify azure file share name in azure file plugin ([#76988](https://github.com/kubernetes/kubernetes/pull/76988), [@andyzhangx](https://github.com/andyzhangx))
* Windows nodes on GCE use a known-working 1809 image rather than the latest 1809 image. ([#76722](https://github.com/kubernetes/kubernetes/pull/76722), [@pjh](https://github.com/pjh))

在这些情况下，你可以使用 **官方 Docker 镜像**，其中包含 **Gunicorn** 作为运行多个 **Uvicorn 工作进程** 的进程管理器，以及一些默认设置来根据当前情况调整工作进程数量 自动CPU核心。 我将在下面的 [Gunicorn - Uvicorn 官方 Docker 镜像](#official-docker-image-with-gunicorn-uvicorn) 中告诉你更多相关信息。

下面一些什么时候这种做法有意义的示例：


#### 一个简单的应用程序

如果你的应用程序**足够简单**，你不需要（至少现在不需要）过多地微调进程数量，并且你可以使用自动默认值，那么你可能需要容器中的进程管理器 （使用官方 Docker 镜像），并且你在**单个服务器**而不是集群上运行它。

---

Common examples of CPU bound operations are things that require complex math processing.

For example:

* **Audio** or **image processing**.
* **Computer vision**: an image is composed of millions of pixels, each pixel has 3 values / colors, processing that normally requires computing something on those pixels, all at the same time.

### Container Images

All container images are available as manifest lists and support the described
architectures. It is also possible to pull a specific architecture directly by
adding the "-$ARCH" suffix  to the container image name.

name | architectures
---- | -------------

                    }
                } else {
                    logger.warn("{} is not available.", generatorName);
                }
            } else if (logger.isDebugEnabled()) {
                logger.debug("No image file exists: {}", noImageFile.getAbsolutePath());
            }
        } catch (final Exception e) {
            logger.warn("Failed to create thumbnail for {}", entity, e);
        }
    }

.fade),this._config.class&&e.addClass(this._config.class),this._config.delay&&500!=this._config.delay&&e.data("delay",this._config.delay);var t=n.default('<div class="toast-header">');if(null!=this._config.image){var a=n.default("<img />").addClass("rounded mr-2").attr("src",this._config.image).attr("alt",this._config.imageAlt);null!=this._config.imageHeight&&a.height(this._config.imageHeight).width("auto"),t.append(a)}if(null!=this._config.icon&&t.append(n.default("<i />").addClass("mr-2").addC...

    to fix this.
-   Once the internal tests pass, we go ahead and merge the code internally as
    well as externally on GitHub.

In a graphical form, the entire lifetime of a PR looks like

![image](https://github.com/tensorflow/tensorflow/assets/52792999/3eea4ca5-daa0-4570-b0b5-2a2b03a724a3)

### Contributor License Agreements

        // local and anonymous classes are prefixed with number (1,2,3...), anonymous classes are
        // entirely numeric whereas local classes have the user supplied name as a suffix
        return CharMatcher.inRange('0', '9').trimLeadingFrom(innerClassName);
      }
      String packageName = getPackageName();
      if (packageName.isEmpty()) {
        return className;
      }

     Material is a musical work, performance, or sound recording,
     Adapted Material is always produced where the Licensed Material is
     synched in timed relation with a moving image.

  b. Adapter's License means the license You apply to Your Copyright
     and Similar Rights in Your contributions to Adapted Material in
     accordance with the terms and conditions of this Public License.