但是将一个项目构建成image的挑战有很多: 依赖、环境变量、绝对路径、网路配置、打包内容.

但是将一个项目构建成image的挑战有很多: 依赖、环境变量、绝对路径、网路配置.

比如我具体遇到的报错:

1. 依赖问题: 有一个python package使用的是github安装而不是pypi的包, 所以需要自定义一下conda 导出的依赖

2. 环境变量问题: 后台需要读取环境变量中的文件, 如token、api-url, 但是docker image的环境变量设置不正确

3. 绝对路径: tensor_path, image_path,使用了绝对路径和相对路径, 在docker中不一致, 找不到文件

4. 网络问题: 容器间的网络连接, mongo、redis的host路径不能使用localhost得使用container_name. 容器与宿主机的网络: api监听失败, 容器中的api需要监听所有ip,, 不能监听localhost, 否则端口转发也不会成功.

# Use NVIDIA's CUDA image as a base
FROM nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu20.04

# Set the working directory
WORKDIR /app

# 设置 DEBIAN_FRONTEND 环境变量，避免交互
ENV DEBIAN_FRONTEND=noninteractive

# 安装 tzdata 时自动选择时区，避免手动交互
RUN ln -fs /usr/share/zoneinfo/America/New_York /etc/localtime && \
    apt update &&\
    apt-get install -y tzdata && \
    dpkg-reconfigure --frontend noninteractive tzdata
# Install necessary dependencies for conda
RUN apt-get update && apt-get install -y \
    wget \
    bzip2 \
    ca-certificates \
    libglib2.0-0 \
    libxext6 \
    libsm6 \
    libxrender1 \
    git \
    && rm -rf /var/lib/apt/lists/*

# Install Miniconda
RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /tmp/miniconda.sh && \
    bash /tmp/miniconda.sh -b -p /opt/conda && \
    rm /tmp/miniconda.sh
ENV PATH=/opt/conda/bin:$PATH

# Copy the environment file and install dependencies
COPY meme-tg-environment.yml .
RUN conda env create -f meme-tg-environment.yml

# Activate the environment
RUN echo "source activate meme" >> ~/.bashrc
ENV PATH /opt/conda/envs/meme/bin:$PATH
# 定义构建时的参数
ARG TOKEN
ARG API_URL
ENV TOKEN=$TOKEN
ENV API_URL=$API_URL
RUN echo "export TOKEN=$TOKEN" >> ~/.bashrc && \
    echo "export API_URL=$API_URL" >> ~/.bashrc

# Set the working directory to /app
WORKDIR /app

# Install any other project files
COPY . /app
RUN mkdir -p /root/.cache/huggingface/


RUN echo '#!/bin/bash\n\
cd /app/memeapi && python celery_worker.py &\n\
cd /app/memeapi && python main.py &\n\
cd /app/imbot && python main.py &\n\
wait' > /app/start_services.sh

# 给脚本可执行权限
RUN chmod +x /app/start_services.sh

# 容器启动时执行脚本
CMD ["/bin/bash", "/app/start_services.sh"]

sudo docker network create meme_network
sudo docker network disconnect bridge 0f6606a9fd35
sudo docker network connect meme_network 0f6606a9fd35
sudo docker network disconnect bridge 04b3f9f0de9e
sudo docker network connect meme_network 04b3f9f0de9e


sudo docker build --build-arg TOKEN=$TOKEN --build-arg API_URL=$API_URL -t meme-cuda-image .
sudo docker run --gpus all -td --network meme_network -p 8000:8000 -v /home/ubuntu/.cache/huggingface/:/root/.cache/huggingface/ --name aibot-meme meme-cuda-image