Docker配置

 {
  "builder": {
    "features": {
      "buildkit": true
    },
    "gc": {
      "defaultKeepStorage": "20GB",
      "enabled": true
    }
  },
  "experimental": false,
  
  "registry-mirrors": [
    "https://docker.m.daocloud.io",
    "https://dockerproxy.com",
    "https://docker.mirrors.ustc.edu.cn",
    "https://docker.nju.edu.cn"
  ],
  "dns": ["223.5.5.5"]
}

Dockerfile

# 基础镜像
FROM nvidia/cuda:latest
# 取消交互操作
ARG DEBIAN_FRONTEND=noninteractive

# cuda key
RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub && \
    apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/7fa2af80.pub

# 切换源
RUN sed -i s@/archive.ubuntu.com/@/mirrors.aliyun.com/@g /etc/apt/sources.list && apt-get clean && apt-get update

# 安装服务
RUN apt-get install -y openssh-server ca-certificates vim curl wget net-tools iproute2 iputils-ping zsh git ntpdate

# oh my zsh
RUN sh -c "$(curl -fsSL https://raw.githubusercontent.com/ohmyzsh/ohmyzsh/master/tools/install.sh)"

# conda
RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh
RUN bash miniconda.sh -b -p /opt/conda
RUN rm miniconda.sh
ENV PATH="/opt/conda/bin:${PATH}"
RUN zsh && conda init --all && pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
ADD .condarc . 

#设置时区
RUN echo "TZ='Asia/Shanghai'; export TZ" >> ~/.profile

# 设置SSH登录密码
RUN echo 'root:miniconda3' | chpasswd
RUN echo 'Port 22\nPermitRootLogin yes\nPasswordAuthentication yes' >> /etc/ssh/sshd_config

# 声明容器会监听的端口
EXPOSE 22

# 启动容器时运行SSH服务
ENTRYPOINT service ssh start && /bin/zsh

Build

$ docker build -t nvidia-cuda-conda-zsh .

启动

$ docker run -itd --gpus=all --shm-size 16G -p 9099:22 -p 6066:6066 -p 7077:7077 --name cuda-conda-zsh nvidia-cuda-conda-zsh /bin/zsh
12edc8ec41ea6311383d65f5b4879edf4d18bc0cdc54329cc738d84fd899fd0a
$ docker exec -it 12edc8ec41ea6311383d65f5b4879edf4d18bc0cdc54329cc738d84fd899fd0a /bin/zsh
➜  /
$ ssh root@127.0.0.1 -p 9099  # 宿主机执行