Centos8-x86_64上安装使用nvidia-docker

一、说明

由于一些原因，遇到了在centos8上使用nvidia-docker，以前基本上没使用过，将自己的安装、配置、使用过程记录一下。

二、步骤

#1.卸载已有组件
yum remove docker \
  docker-client \
  docker-client-latest \
  docker-common \
  docker-latest \
  docker-latest-logrotate \
  docker-logrotate \
  docker-engine

#2.安装基础组件
yum install -y yum-utils \
  device-mapper-persistent-data \
  lvm2

#3.配置仓库
yum-config-manager \
  --add-repo \
  https://mirrors.aliyun.com/docker-ce/linux/centos/docker-ce.repo

#4.安装docker-ce*
yum list docker-ce --showduplicates | sort -r  #查看版本
#安装最新版本
yum install docker-ce docker-ce-cli containerd.io
#如果要安装指定版本
yum install docker-ce-<VERSION_STRING> docker-ce-cli-<VERSION_STRING> containerd.io

#5.启动docker并在不使用GPU时运行容器
systemctl start docker
docker run hello-world 或 docker run swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/library/hello-world:latest

#6.安装 nvidia-container-* 包
yum install -y nvidia-container-toolkit install nvidia-container-runtime

#7.Docker 守护进程以识别 NVIDIA 容器运行时（关键）
nvidia-ctk runtime configure --runtime=docker

#8.修改/etc/docker/daemon.json文件，添加默认运行时为nvidia
vi /etc/docker/daemon.json
{
    "exec-opts": ["native.cgroupdriver=systemd"],
    "dns": [
        "223.5.5.5",
        "223.6.6.6",
        "202.103.0.68",
        "8.8.8.8"
    ],
    "log-opts": {
        "max-file": "5",
        "max-size": "50m"
    },
    "registry-mirrors": [
        "https://hub.rat.dev",
        "https://registry.aliyuncs.com",
        "https://registry.docker-cn.com",
        "https://docker.chenby.cn",
        "https://docker.registry.cyou",
        "https://docker-cf.registry.cyou",
        "https://dockercf.jsdelivr.fyi",
        "https://docker.jsdelivr.fyi",
        "https://dockertest.jsdelivr.fyi",
        "https://dockerproxy.com",
        "https://docker.m.daocloud.io",
        "https://docker.nju.edu.cn",
        "https://docker.mirrors.sjtug.sjtu.edu.cn",
        "https://docker.mirrors.ustc.edu.cn",
        "https://mirror.iscas.ac.cn",
        "https://docker.rainbond.cc"
    ],
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "/usr/local/nvidia/toolkit/nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

#9.加载daemon.json文件并重启docker服务
systemctl daemon-reload
systemctl restart docker

#10.创建带NVIDIA GPU的容器
docker run -itd \
  --name ubuntu2204 \
  --gpus='"device=0,1"' \
  --restart=always \
  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/library/ubuntu:22.04
#如果要使用全部NVIDIA GPU
docker run -itd \
  --name ubuntu2204 \
  --gpus all \
  --restart=always \
  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/library/ubuntu:22.04