Docker 中的多阶段构建显示未知路径

Question

我想构建一个多阶段构建的 Docker 镜像。该图像基于 FastAPI 和 Uvicorn。

如果我使用以下

Dockerfile

进行多阶段构建

# Stage 1: Build stage
FROM pytorch/pytorch:2.1.2-cuda12.1-cudnn8-runtime as builder

# set working directory
WORKDIR /usr/src

# set environment variables
ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1

RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 python3-pip git \
    && ln -sf python3 /usr/bin/python \
    && ln -sf pip3 /usr/bin/pip \
    && pip install --upgrade pip

# install python dependencies
RUN pip install --upgrade pip setuptools wheel
COPY ./requirements.txt .
RUN pip install -r requirements.txt

RUN apt-get remove -y git


# Stage 2: Final stage
FROM pytorch/pytorch:2.1.2-cuda12.1-cudnn8-runtime

# set environment variables
ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1

RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 \
    && ln -sf python3 /usr/bin/python

# Copy the installed dependencies from the previous stage
COPY --from=builder /usr/local /usr/local
ENV PATH=/usr/local/bin:$PATH

WORKDIR /usr/src

# Copy your app code
COPY ./app/ ./app/

并使用以下命令运行它

docker-compose.yml

version: '3.8'

services:

  scraiber_gpt:
    build:
      context: .
      dockerfile: Dockerfile
      secrets:
        - githubpat
    command: uvicorn app.main:app --reload --workers 1 --host 0.0.0.0 --port 8000
    volumes:
      - ./src:/usr/src
    ports:
      - '8000:8000'

我明白了

Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: exec: "uvicorn": executable file not found in $PATH: unknown

但是，对于使用

Dockerfile

的非多阶段构建

FROM pytorch/pytorch:2.1.2-cuda12.1-cudnn8-runtime

# set working directory
WORKDIR /usr/src

# set environment variables
ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1

RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 python3-pip git \
    && ln -sf python3 /usr/bin/python \
    && ln -sf pip3 /usr/bin/pip \
    && pip install --upgrade pip

# install python dependencies
RUN pip install --upgrade pip setuptools wheel
COPY ./requirements.txt .
RUN pip install -r requirements.txt

RUN apt-get remove -y git

# Copy your app code
COPY ./app/ ./app/

事情往往会起作用。因此，我想知道如何使多阶段构建工作。

Answer 1

最好使用 virtualenv 来管理你的依赖项，试试这个：

# Stage 1: Build stage
FROM pytorch/pytorch:2.1.2-cuda12.1-cudnn8-runtime as builder

# set working directory
WORKDIR /usr/src

# set environment variables
ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1




RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 python3-pip git \
    && ln -sf python3 /usr/bin/python \
    && ln -sf pip3 /usr/bin/pip \
    && pip install --upgrade pip

RUN python -m venv /opt/venv
ENV PATH="/opt/venv/bin:$PATH"

# install python dependencies
RUN pip install --upgrade pip setuptools wheel
COPY ./requirements.txt .
RUN pip install -r requirements.txt

RUN apt-get remove -y git


# Stage 2: Final stage
FROM pytorch/pytorch:2.1.2-cuda12.1-cudnn8-runtime

# set environment variables
ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1

RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 \
    && ln -sf python3 /usr/bin/python

NV PYTHONUNBUFFERED 1

RUN apt-get update \
    && apt-get install -y --no-install-recommends python3 \
    && ln -sf python3 /usr/bin/python

COPY --from=builder /opt/venv /opt/venv

WORKDIR /usr/src

# Copy your app code
COPY ./app/ ./app/