docker-build

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Docker Build for Python/FastAPI Applications

Python/FastAPI应用的Docker构建

Overview

概述

Containerize Python and FastAPI applications using Docker with production-grade multi-stage builds, layer caching, security hardening, and Docker Compose orchestration. All patterns target Python 3.12+, use

uv

for dependency management, and follow the principle of minimal, reproducible images.

Refer to the asset templates bundled with this skill for ready-to-use starting points:

```
assets/Dockerfile.fastapi
```
-- production multi-stage Dockerfile
```
assets/Dockerfile.dev
```
-- development Dockerfile with hot reload
```
assets/docker-compose.yml
```
-- Compose stack with PostgreSQL and Redis
```
assets/.dockerignore
```
-- ignore rules for lean build contexts

使用Docker对Python和FastAPI应用进行容器化，采用生产级别的多阶段构建、层缓存、安全加固以及Docker Compose编排。所有模式均针对Python 3.12+版本，使用

uv

进行依赖管理，并遵循最小化、可复现镜像的原则。

参考此技能附带的资产模板，获取可直接使用的起点：

```
assets/Dockerfile.fastapi
```
-- 生产环境多阶段Dockerfile
```
assets/Dockerfile.dev
```
-- 带热重载的开发环境Dockerfile
```
assets/docker-compose.yml
```
-- 包含PostgreSQL和Redis的Compose栈
```
assets/.dockerignore
```
-- 精简构建上下文的忽略规则

Base Image Selection

基础镜像选择

Choose the base image according to the deployment target and dependency requirements.

根据部署目标和依赖要求选择基础镜像。

python:3.12-slim (recommended)

python:3.12-slim（推荐）

The default choice for most FastAPI projects. Based on Debian Bookworm with a minimal package set. Binary wheels from PyPI install without issues, and system libraries such as

libpq

can be added via

apt-get

dockerfile

FROM python:3.12-slim AS base

大多数FastAPI项目的默认选择。基于Debian Bookworm，包含最小化的软件包集合。PyPI的二进制wheel可顺利安装，可通过

apt-get

添加

libpq

等系统库。

dockerfile

FROM python:3.12-slim AS base

python:3.12-alpine

Smaller download size, but Alpine uses musl libc. Packages with C extensions (e.g.,

psycopg2

numpy

pandas

) often fail to install or require building from source, negating the size advantage. Avoid unless the dependency tree is pure Python.

下载体积更小，但Alpine使用musl libc。带有C扩展的软件包（如

psycopg2

、

numpy

、

pandas

）通常安装失败或需要从源码构建，抵消了体积优势。除非依赖树全部为纯Python，否则避免使用。

ghcr.io/astral-sh/uv:python3.12-bookworm-slim

Ships with

uv

pre-installed. Useful when the build should not fetch

uv

at build time. The image is Debian-based and behaves like

python:3.12-slim

otherwise.

dockerfile

FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim AS builder

预安装了

uv

。适用于构建时无需拉取

uv

的场景。该镜像基于Debian，其他行为与

python:3.12-slim

一致。

dockerfile

FROM ghcr.io/astral-sh/uv:python3.12-bookworm-slim AS builder

Multi-Stage Build Pattern

多阶段构建模式

Separate dependency installation from the final runtime image to keep the production image small and free of build tools.

将依赖安装与最终运行时镜像分离，以保持生产镜像体积小巧且不含构建工具。

Stage 1 -- Builder

阶段1 -- 构建器

Install

uv

, copy only the dependency manifests, and resolve/install dependencies into a virtual environment. This stage may contain compilers and header packages that must not ship in production.

dockerfile

FROM python:3.12-slim AS builder

COPY --from=ghcr.io/astral-sh/uv /uv /usr/local/bin/uv

ENV UV_COMPILE_BYTECODE=1
ENV UV_LINK_MODE=copy

WORKDIR /app

COPY pyproject.toml uv.lock ./
RUN uv sync --frozen --no-dev --no-install-project

COPY . .
RUN uv sync --frozen --no-dev

安装

uv

，仅复制依赖清单，并将依赖解析/安装到虚拟环境中。此阶段可包含编译器和头文件包，但这些不得随生产镜像发布。

dockerfile

FROM python:3.12-slim AS builder

COPY --from=ghcr.io/astral-sh/uv /uv /usr/local/bin/uv

ENV UV_COMPILE_BYTECODE=1
ENV UV_LINK_MODE=copy

WORKDIR /app

COPY pyproject.toml uv.lock ./
RUN uv sync --frozen --no-dev --no-install-project

COPY . .
RUN uv sync --frozen --no-dev

Stage 2 -- Runtime

阶段2 -- 运行时

Start from a clean slim image, copy only the virtual environment and application source from the builder, create a non-root user, and declare the entrypoint.

dockerfile

FROM python:3.12-slim AS runtime

WORKDIR /app

COPY --from=builder /app/.venv /app/.venv
COPY --from=builder /app/src /app/src

ENV PATH="/app/.venv/bin:$PATH"

RUN useradd --create-home --shell /bin/bash appuser
USER appuser

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

从干净的slim镜像开始，仅从构建器复制虚拟环境和应用源码，创建非root用户，并声明入口点。

dockerfile

FROM python:3.12-slim AS runtime

WORKDIR /app

COPY --from=builder /app/.venv /app/.venv
COPY --from=builder /app/src /app/src

ENV PATH="/app/.venv/bin:$PATH"

RUN useradd --create-home --shell /bin/bash appuser
USER appuser

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

uv Integration in Docker

Docker中的uv集成

Installing uv

安装uv

The fastest method is a static binary copy from the official image:

dockerfile

COPY --from=ghcr.io/astral-sh/uv /uv /usr/local/bin/uv

Alternatively, install via pip (slower, but works on any base):

dockerfile

RUN pip install --no-cache-dir uv

最快的方法是从官方镜像复制静态二进制文件：

dockerfile

COPY --from=ghcr.io/astral-sh/uv /uv /usr/local/bin/uv

或者通过pip安装（速度较慢，但可在任何基础镜像上运行）：

dockerfile

RUN pip install --no-cache-dir uv

Dependency Resolution

依赖解析

Always pass

--frozen

uv sync

inside Docker so that the lockfile is used as-is without resolution. This guarantees reproducible builds.

dockerfile

RUN uv sync --frozen --no-dev

For uv dependency management outside of Docker, consult the

uv

skill.

在Docker内部执行

uv sync

时，始终传递

--frozen

参数，以便直接使用锁定文件而不进行重新解析。这可确保构建的可复现性。

dockerfile

RUN uv sync --frozen --no-dev

关于Docker之外的uv依赖管理，请参考

uv

技能文档。

Layer Ordering for Cache Efficiency

缓存效率的层顺序

Docker caches each layer. When a layer changes, every subsequent layer is rebuilt. Order instructions from least to most frequently changing:

Base image and system packages
Copy dependency manifests (
```
pyproject.toml
```
,
```
uv.lock
```
)
Install dependencies (
```
uv sync
```
)
Copy application source code
Any final build steps

dockerfile

undefined

Docker会缓存每个层。当某一层发生变化时，后续所有层都将重新构建。按变更频率从低到高排列指令：

基础镜像和系统软件包
复制依赖清单（
```
pyproject.toml
```
、
```
uv.lock
```
）
安装依赖（
```
uv sync
```
）
复制应用源码
任何最终构建步骤

dockerfile

undefined

Step 2 -- manifests change infrequently

步骤2 -- 清单变更频率低

COPY pyproject.toml uv.lock ./

Step 3 -- re-runs only when manifests change

步骤3 -- 仅在清单变更时重新运行

RUN uv sync --frozen --no-dev --no-install-project

Step 4 -- changes on every code edit

步骤4 -- 每次代码编辑都会变更

COPY src/ src/


This ordering ensures that dependency installation is cached across most code
changes, drastically reducing rebuild times.

---

COPY src/ src/


此顺序确保依赖安装在大多数代码变更时都能被缓存，大幅缩短重建时间。

---

Security Hardening

安全加固

Non-Root User

非Root用户

Never run the application as root inside the container. Create a dedicated user and switch to it before

CMD

dockerfile

RUN useradd --create-home --shell /bin/bash appuser
USER appuser

For stricter security, use a numeric UID and no login shell:

dockerfile

RUN adduser --system --uid 1001 --no-create-home appuser
USER 1001

永远不要在容器内以root身份运行应用。创建专用用户，并在

CMD

之前切换到该用户。

dockerfile

RUN useradd --create-home --shell /bin/bash appuser
USER appuser

如需更严格的安全设置，可使用数字UID且不设置登录shell：

dockerfile

RUN adduser --system --uid 1001 --no-create-home appuser
USER 1001

Minimal Packages

最小化软件包

Do not install editors, debug tools, or documentation packages in the production image. If build-time packages are necessary (e.g.,

gcc

libpq-dev

), install them only in the builder stage.

不要在生产镜像中安装编辑器、调试工具或文档软件包。如果构建时需要某些软件包（如

gcc

、

libpq-dev

），仅在构建器阶段安装它们。

.dockerignore

Exclude files that must not enter the build context: version control metadata, local environment files, test suites, caches, and documentation. See

assets/.dockerignore

for a comprehensive template.

排除不得进入构建上下文的文件：版本控制元数据、本地环境文件、测试套件、缓存和文档。参考

assets/.dockerignore

获取全面的模板。

Pin Image Digests in CI

在CI中固定镜像摘要

For reproducible CI builds, pin the base image to a digest:

dockerfile

FROM python:3.12-slim@sha256:<digest> AS builder

为实现可复现的CI构建，将基础镜像固定到具体的摘要值：

dockerfile

FROM python:3.12-slim@sha256:<digest> AS builder

HEALTHCHECK Instruction

HEALTHCHECK指令

Define a health check so orchestrators (Docker Swarm, Compose, ECS) can detect unresponsive containers and restart them automatically.

dockerfile

HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD ["python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"] || exit 1

curl

is available in the runtime image, prefer the simpler form:

dockerfile

HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD curl -f http://localhost:8000/health || exit 1

Ensure the FastAPI application exposes a lightweight

/health

endpoint that returns HTTP 200.

定义健康检查，以便编排器（Docker Swarm、Compose、ECS）能够检测无响应的容器并自动重启。

dockerfile

HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD ["python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8000/health')"] || exit 1

如果运行时镜像中包含

curl

，可使用更简洁的形式：

dockerfile

HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
  CMD curl -f http://localhost:8000/health || exit 1

确保FastAPI应用暴露一个轻量级的

/health

端点，返回HTTP 200状态码。

Environment Variables

环境变量

Set these early in the Dockerfile to influence Python runtime behavior:

dockerfile

undefined

在Dockerfile中尽早设置这些变量，以影响Python运行时行为：

dockerfile

undefined

Send stdout/stderr straight to the terminal without buffering

将stdout/stderr直接发送到终端，不进行缓冲

ENV PYTHONUNBUFFERED=1

Prevent .pyc file creation inside the container

防止在容器内创建.pyc文件

ENV PYTHONDONTWRITEBYTECODE=1


Application-specific variables (database URL, secret keys) should be injected
at runtime via `docker run --env-file` or the Compose `env_file` directive,
never baked into the image.

---

ENV PYTHONDONTWRITEBYTECODE=1


应用特定的变量（数据库URL、密钥）应在运行时通过`docker run --env-file`或Compose的`env_file`指令注入，永远不要嵌入到镜像中。

---

CMD with Exec Form

执行格式的CMD

Always use the exec form (JSON array) for

CMD

so that

uvicorn

receives signals directly from Docker and can shut down gracefully.

dockerfile

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

Avoid the shell form (

CMD uvicorn ...

) because it wraps the process in

/bin/sh -c

, which swallows

SIGTERM

and delays container stops.

For uvicorn production configuration (workers, timeouts, keep-alive), consult the

uvicorn

skill.

始终使用执行格式（JSON数组）编写

CMD

，以便

uvicorn

直接接收来自Docker的信号并优雅关闭。

dockerfile

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"]

避免使用shell格式（

CMD uvicorn ...

），因为它会将进程包装在

/bin/sh -c

中，这会吞噬

SIGTERM

信号并延迟容器停止。

关于uvicorn的生产配置（工作进程、超时、保持连接），请参考

uvicorn

技能文档。

Docker Compose

Use Docker Compose to orchestrate the application alongside backing services such as PostgreSQL and Redis. See

assets/docker-compose.yml

for a complete template.

使用Docker Compose将应用与PostgreSQL、Redis等后端服务一起编排。参考

assets/docker-compose.yml

获取完整模板。

Service Definitions

服务定义

yaml

services:
  app:
    build:
      context: .
      dockerfile: Dockerfile.fastapi
    ports:
      - "8000:8000"
    env_file:
      - .env
    depends_on:
      db:
        condition: service_healthy
      redis:
        condition: service_healthy

yaml

services:
  app:
    build:
      context: .
      dockerfile: Dockerfile.fastapi
    ports:
      - "8000:8000"
    env_file:
      - .env
    depends_on:
      db:
        condition: service_healthy
      redis:
        condition: service_healthy

Health Checks in Compose

Compose中的健康检查

Each service should declare a

healthcheck

so that

depends_on

with

condition: service_healthy

works correctly.

yaml

db:
  image: postgres:16-alpine
  healthcheck:
    test: ["CMD-SHELL", "pg_isready -U postgres"]
    interval: 10s
    timeout: 5s
    retries: 5

每个服务都应声明

healthcheck

，以便

depends_on

配合

condition: service_healthy

正常工作。

yaml

db:
  image: postgres:16-alpine
  healthcheck:
    test: ["CMD-SHELL", "pg_isready -U postgres"]
    interval: 10s
    timeout: 5s
    retries: 5

Volumes

卷

Persist database data across container recreations:

yaml

volumes:
  postgres_data:
  redis_data:

在容器重新创建时持久化数据库数据：

yaml

volumes:
  postgres_data:
  redis_data:

Networks

网络

Isolate inter-service traffic on a dedicated bridge network:

yaml

networks:
  backend:
    driver: bridge

在专用的桥接网络上隔离服务间通信：

yaml

networks:
  backend:
    driver: bridge

Development vs Production Dockerfiles

开发与生产Dockerfile对比

Maintain separate Dockerfiles for each environment.

为每个环境维护单独的Dockerfile。

Production (

Dockerfile.fastapi

)

生产环境（

Dockerfile.fastapi

）

Multi-stage build
```
--no-dev
```
dependencies only
Source code copied into the image
Non-root user, health check, exec-form CMD

多阶段构建
仅包含
```
--no-dev
```
依赖
源码复制到镜像中
非root用户、健康检查、执行格式的CMD

Development (

Dockerfile.dev

)

开发环境（

Dockerfile.dev

）

Single stage for simplicity
All dependencies including dev extras (
```
pytest
```
,
```
ruff
```
, etc.)
Source code mounted via a bind volume -- do not
```
COPY
```
source
```
--reload
```
flag on uvicorn for hot reload

dockerfile

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000", "--reload"]

Override the production Compose file with a development variant:

yaml

undefined

单阶段构建，简化流程
包含所有依赖，包括开发附加组件（
```
pytest
```
、
```
ruff
```
等）
源码通过绑定卷挂载 -- 不要
```
COPY
```
源码
uvicorn使用
```
--reload
```
标志实现热重载

dockerfile

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000", "--reload"]

使用开发变体覆盖生产Compose文件：

yaml

undefined

docker-compose.override.yml

services: app: build: dockerfile: Dockerfile.dev volumes: - .:/app

---

services: app: build: dockerfile: Dockerfile.dev volumes: - .:/app

---

Build Arguments

构建参数

Use

ARG

to parameterize the build without creating multiple Dockerfiles.

dockerfile

ARG PYTHON_VERSION=3.12
FROM python:${PYTHON_VERSION}-slim AS builder

ARG APP_VERSION=0.0.0
LABEL org.opencontainers.image.version=${APP_VERSION}

Pass values at build time:

bash

docker build --build-arg PYTHON_VERSION=3.13 --build-arg APP_VERSION=1.2.0 .

使用

ARG

参数化构建，无需创建多个Dockerfile。

dockerfile

ARG PYTHON_VERSION=3.12
FROM python:${PYTHON_VERSION}-slim AS builder

ARG APP_VERSION=0.0.0
LABEL org.opencontainers.image.version=${APP_VERSION}

在构建时传递参数值：

bash

docker build --build-arg PYTHON_VERSION=3.13 --build-arg APP_VERSION=1.2.0 .

Volume Mounts for Development

开发环境的卷挂载

Bind-mount the project directory into the container so that file changes on the host are immediately visible inside the container, enabling uvicorn's

--reload

watcher.

bash

docker run -v "$(pwd)":/app -p 8000:8000 myapp-dev

In Compose:

yaml

services:
  app:
    volumes:
      - .:/app

Avoid bind-mounting over the

.venv

directory. If the virtual environment lives inside the project tree, use an anonymous volume to shadow it:

yaml

volumes:
  - .:/app
  - /app/.venv

将项目目录绑定挂载到容器中，以便主机上的文件变更能立即在容器内生效，启用uvicorn的

--reload

监视器。

bash

docker run -v "$(pwd)":/app -p 8000:8000 myapp-dev

在Compose中：

yaml

services:
  app:
    volumes:
      - .:/app

避免绑定挂载覆盖

.venv

目录。如果虚拟环境位于项目目录内，使用匿名卷将其隐藏：

yaml

volumes:
  - .:/app
  - /app/.venv

Image Size Reduction Checklist

镜像体积缩减清单

Use
```
python:3.12-slim
```
instead of the full
```
python
```
image.
Apply multi-stage builds; keep compilers in the builder stage only.
Combine
```
RUN
```
commands where logical to reduce layer count.
Pass
```
--no-cache-dir
```
to pip or use
```
uv
```
(which never caches by default).

Remove apt lists after installing system packages:

RUN apt-get update && apt-get install -y --no-install-recommends pkg && rm -rf /var/lib/apt/lists/*

Add a thorough
```
.dockerignore
```
to minimize the build context.
Compile bytecode at build time (
```
UV_COMPILE_BYTECODE=1
```
) and skip
```
.pyc
```
generation at runtime (
```
PYTHONDONTWRITEBYTECODE=1
```
).

使用
```
python:3.12-slim
```
而非完整的
```
python
```
镜像。
采用多阶段构建；仅在构建器阶段保留编译器。
在合理的情况下合并
```
RUN
```
命令，以减少层数。
向pip传递
```
--no-cache-dir
```
参数，或使用
```
uv
```
（默认从不缓存）。

安装系统软件包后删除apt列表：

RUN apt-get update && apt-get install -y --no-install-recommends pkg && rm -rf /var/lib/apt/lists/*

添加完善的
```
.dockerignore
```
以最小化构建上下文。
在构建时编译字节码（
```
UV_COMPILE_BYTECODE=1
```
），并在运行时跳过
```
.pyc
```
生成（
```
PYTHONDONTWRITEBYTECODE=1
```
）。

Multi-Architecture Builds

多架构构建

Build images for multiple platforms using

docker buildx

bash

undefined

使用

docker buildx

为多个平台构建镜像：

bash

undefined

Create a buildx builder

创建buildx构建器

docker buildx create --name multiarch --use

Build and push for amd64 and arm64

为amd64和arm64构建并推送

docker buildx build --platform linux/amd64,linux/arm64 -t myapp:latest --push .


When using multi-arch builds, ensure all base images support the target platforms. Avoid architecture-specific binaries in `COPY` instructions.

---

docker buildx build --platform linux/amd64,linux/arm64 -t myapp:latest --push .


使用多架构构建时，确保所有基础镜像支持目标平台。避免在`COPY`指令中使用特定架构的二进制文件。

---

BuildKit Secrets

BuildKit密钥

Mount secrets at build time without baking them into image layers:

dockerfile

undefined

在构建时挂载密钥，而不将其嵌入到镜像层中：

dockerfile

undefined

syntax=docker/dockerfile:1

RUN --mount=type=secret,id=pip_index_url
PIP_INDEX_URL=$(cat /run/secrets/pip_index_url)
uv sync --frozen --no-dev


Pass the secret at build time:

```bash
docker build --secret id=pip_index_url,env=PIP_INDEX_URL .

Secrets are available only during the

RUN

instruction and never persist in the image history.

RUN --mount=type=secret,id=pip_index_url
PIP_INDEX_URL=$(cat /run/secrets/pip_index_url)
uv sync --frozen --no-dev


在构建时传递密钥：

```bash
docker build --secret id=pip_index_url,env=PIP_INDEX_URL .

密钥仅在

RUN

指令期间可用，永远不会保留在镜像历史中。

Common Labels

通用标签

Apply OCI-standard labels for image metadata:

dockerfile

LABEL org.opencontainers.image.title="my-fastapi-app"
LABEL org.opencontainers.image.description="FastAPI application"
LABEL org.opencontainers.image.version="1.0.0"
LABEL org.opencontainers.image.source="https://github.com/org/repo"

应用OCI标准标签以添加镜像元数据：

dockerfile

LABEL org.opencontainers.image.title="my-fastapi-app"
LABEL org.opencontainers.image.description="FastAPI application"
LABEL org.opencontainers.image.version="1.0.0"
LABEL org.opencontainers.image.source="https://github.com/org/repo"

Cross-References

交叉引用

For uvicorn production configuration (workers, timeouts, logging), consult the
```
uvicorn
```
skill.
For uv dependency management (lockfiles, workspaces, scripts), consult the
```
uv
```
skill.

关于uvicorn的生产配置（工作进程、超时、日志），请参考
```
uvicorn
```
技能文档。
关于uv依赖管理（锁定文件、工作区、脚本），请参考
```
uv
```
技能文档。

docker-build

Original

Translation

Docker Build for Python/FastAPI Applications

Python/FastAPI应用的Docker构建

Overview

概述

Base Image Selection

基础镜像选择

python:3.12-slim (recommended)

python:3.12-slim（推荐）

python:3.12-alpine

python:3.12-alpine

ghcr.io/astral-sh/uv:python3.12-bookworm-slim

ghcr.io/astral-sh/uv:python3.12-bookworm-slim

Multi-Stage Build Pattern

多阶段构建模式

Stage 1 -- Builder

阶段1 -- 构建器

Stage 2 -- Runtime

阶段2 -- 运行时

uv Integration in Docker

Docker中的uv集成

Installing uv

安装uv

Dependency Resolution

依赖解析

Layer Ordering for Cache Efficiency

缓存效率的层顺序

Step 2 -- manifests change infrequently

步骤2 -- 清单变更频率低

Step 3 -- re-runs only when manifests change

步骤3 -- 仅在清单变更时重新运行

Step 4 -- changes on every code edit

步骤4 -- 每次代码编辑都会变更

Security Hardening

安全加固

Non-Root User

非Root用户

Minimal Packages

最小化软件包

.dockerignore

.dockerignore

Pin Image Digests in CI

在CI中固定镜像摘要

HEALTHCHECK Instruction

HEALTHCHECK指令

Environment Variables

环境变量

Send stdout/stderr straight to the terminal without buffering

将stdout/stderr直接发送到终端，不进行缓冲

Prevent .pyc file creation inside the container

防止在容器内创建.pyc文件

CMD with Exec Form

执行格式的CMD

Docker Compose

Docker Compose

Service Definitions

服务定义

Health Checks in Compose

Compose中的健康检查

Volumes

卷

Networks

网络

Development vs Production Dockerfiles

开发与生产Dockerfile对比

Production (Dockerfile.fastapi)

生产环境（Dockerfile.fastapi）

Development (Dockerfile.dev)

开发环境（Dockerfile.dev）

docker-compose.override.yml

docker-compose.override.yml

Build Arguments

构建参数

Volume Mounts for Development

开发环境的卷挂载

Image Size Reduction Checklist

镜像体积缩减清单

Multi-Architecture Builds

Production (
`Dockerfile.fastapi`
)

生产环境（
`Dockerfile.fastapi`
）

Development (
`Dockerfile.dev`
)

开发环境（
`Dockerfile.dev`
）