Docker+GPU配置+Python打包部署全流程

确保主机docker环境正常，可以调用GPU

测试命令：

sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi

若不正常，则参考以下步骤解决：

# 安装并确认docker正常，尽量不要用docker-desktop
# 参考 https://docs.docker.com/engine/install/ubuntu/
# 有问题随时重启docker：systemctl daemon-reload 和 systemctl restart docker
sudo apt-get update
sudo apt-get install ca-certificates curl gnupg
 
sudo install -m 0755 -d /etc/apt/keyrings
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /etc/apt/keyrings/docker.gpg
sudo chmod a+r /etc/apt/keyrings/docker.gpg
 
echo \
    "deb [arch="$(dpkg --print-architecture)" signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu \
    "$(. /etc/os-release && echo "$VERSION_CODENAME")" stable" | \
    sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
 
sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
docker --version
sudo docker run hello-word
 
# 为了可以调用GPU，安装NVIDIA Container Toolkit并确认
# 参考 https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
# 参考 https://blog.csdn.net/Tang_Zhe/article/details/122936986
distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
 
sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit
sudo nano /etc/docker/daemon.json 并输入{换行}保存
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker
sudo docker run --rm --runtime=nvidia --gpus all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi

docker出现问题，考虑重装：

sudo apt-get remove docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
sudo apt-get install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

编写dockerfile如下

# v1.2 主要是opengl安装
# 使用基础镜像
FROM ubuntu:20.04
# 避免在安装过程中出现交互式对话框
ARG DEBIAN_FRONTEND=noninteractive
# 设置工作目录
WORKDIR ./ImgEvalSys
# 时区设置，避免奇奇怪怪的问题
ENV TZ=Asia/Shanghai
# 复制项目文件到容器中
ADD . /ImgEvalSys
# Info
LABEL version="1.2"
LABEL manager="Shuai He <hs19951021@bupt.edu.cn>"
LABEL maintainer1="Rui Xie <chenke771@bupt.edu.cn>"
LABEL maintainer2="Yi Xiao <@bupt.edu.cn>"
# apt安装opengl和python3.9
RUN apt-get update && apt-get upgrade -y
RUN apt-get install -y libgl1-mesa-glx libgl1-mesa-dri
RUN apt-get install -y python3.9 python3-pip
RUN update-alternatives --install /usr/bin/python python /usr/bin/python3.9 1 && \
update-alternatives --set python /usr/bin/python3.9
RUN apt-get install libglib2.0-0
# # 安装依赖(为避免问题，在docker创建完成后再安装)
# RUN python3.9 -m pip install -r requirements.txt -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
# RUN python3.9 -m pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn
# RUN python3.9 -m pip install -r requirements.txt
# 替换opencv为headless版本
RUN python3.9 -m pip uninstall opencv-python -y
RUN python3.9 -m pip install opencv-python-headless -i http://pypi.douban.com/simple --trusted-host pypi.douban.com
# 补充安装有问题的依赖包
RUN python3.9 -m pip install opencv-contrib-python-headless
RUN python3.9 -m pip install colour-science
# Debug版启动命令：进入bash命令行
CMD ["/bin/bash"]
# 后续若让container启动时候自动运行测评系统，可用以下替代
# Release版启动命令：直接启动系统
# CMD ["python", "./run.py"]

构建image并启动container（必要时加sudo）

# cd 项目文件夹，构建image
docker build -t imevsys-headless .
 
# 启动container
# 启动到bash（调试模式）
sudo docker run --runtime=nvidia --gpus all -p 9999:9999 -it imevsys-headless
# 或 直接启动系统
sudo docker run --runtime=nvidia --gpus all -p 9999:9999 -it imevsys-headless python web_gui.py
# 或 设定挂载目录
sudo docker run --runtime=nvidia --gpus all -v /home/xr/Templates/ImgEvalSysss:/ImgEvalSys -p 9999:9999 -it imevsys-headless python web_gui.py

导出/载入image

sudo docker save -o ImgEvalSys.tar imevsys-headless
sudo docker load -i ImgEvalSys.tar
sudo chmod 777 ImgEvalSys.tar

依赖发生微小变化后重新打包

# 查看container运行情况，寻找<container_id>
sudo docker ps -a
 
# 在容器内：替换环境
pip install colour-science
 
# 在主机内：提交、打包、更改权限
sudo docker commit <container_id> imevsys-headless:version
sudo docker save -o ImgEvalSys.tar imevsys-headless:version
sudo chmod 777 ./ImgEvalSys.tar
 
# 其他可选系统维护操作：修改image的tag（实际上是新建一个，再删除旧image）
sudo docker image tag imevsys-headless:latest imevsys-headless:pre
sudo docker image rmi imevsys-headless:latest
 
#  如果改不了，就先ps查找再rm删除占用image的container
sudo docker ps -a
sudo docker rm <container_id>
 
# 有时在启动docker container时需要同时运行不止一条命令，这时在command处使用：sh -c ' cmd1 && cmd2 '即可。
sudo docker run --runtime=nvidia --gpus all -v /home/xr/Templates/ImgEvalSys:/ImgEvalSys -p 9999:9999 -it imevsys-headless sh -c 'pip install -r req_0925.txt -i https://pypi.tuna.tsinghua.edu.cn/simple --trusted-host pypi.tuna.tsinghua.edu.cn && python web_gui.py'

Ubuntu conda和pip