1.环境

GPU: `Tesla T4`
OS:`CentOS-7`
Kernel: `3.10.0-1160.105.1.el7.x86_64`

2. 驱动安装

下载地址 https://www.nvidia.cn/Download/index.aspx

./NVIDIA-Linux-x86_64-535.129.03.run 

3.nvidia-docker 安装

• docker

  [root@skywalking-0001 ~]# wget -O /etc/yum.repos.d/CentOS-Base.repo https://mirrors.aliyun.com/repo/Centos-7.repo
  [root@skywalking-0001 ~]# yum install -y yum-utils device-mapper-persistent-data lvm2
  [root@skywalking-0001 ~]# yum-config-manager --add-repo https://mirrors.aliyun.com/docker-ce/linux/centos/docker-ce.repo
  [root@skywalking-0001 ~]# sed -i 's+download.docker.com+mirrors.aliyun.com/docker-ce+' /etc/yum.repos.d/docker-ce.repo
  [root@skywalking-0001 ~]# yum makecache fast
  [root@skywalking-0001 ~]# yum -y install docker-ce-19.03.9-3.el7
  # 配置镜像加速器
  [root@skywalking-0001 ~]# setenforce 0  && systemctl stop firewalld
  [root@skywalking-0001 ~]# mkdir -p /etc/docker
  [root@skywalking-0001 ~]# tee /etc/docker/daemon.json <<-'EOF'
  {
    "registry-mirrors": ["https://6ysopw9t.mirror.aliyuncs.com"]
  }
  EOF
  # 修改docker 工作目录
  [root@skywalking-0001 ~]# mkdir /data/docker
  # 在ExecStart=/usr/bin/dockerd -H fd:// --containerd=/run/containerd/containerd.sock 后追加 --data-root=/data/docker
  [root@skywalking-0001 ~]# vim /usr/lib/systemd/system/docker.service
  
  [root@skywalking-0001 ~]# systemctl daemon-reload && systemctl restart docker
  

• Nvidia-docker

  distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
  curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.repo | sudo tee /etc/yum.repos.d/nvidia-docker.repo
  
  sudo yum install -y nvidia-container-toolkit
  sudo yum install -y nvidia-docker2
  sudo systemctl restart docker
  

• 验证

  docker run --gpus all nvidia/cuda:10.0-base /bin/sh -c "while true; do echo hello world; sleep 1; done"
  docker run --help | grep -i gpus
  

4.常见问题

• Unable to load: nvidia-installer ncurses v6 user interface

  需要禁用nouveau

  临时禁用：
     modprobe -r nouveau
  永久禁用：（需要重启）
    mv /etc/modprobe.d/nvidia-installer-disable-nouveau.conf /etc/modprobe.d/blacklist.conf
    mv /boot/initramfs-$(uname -r).img  /boot/initramfs-3.10.0-1160.el7.x86_64.img.bak
    dracut -v  /boot/initramfs-$(uname -r).img $(uname -r)
    reboot
  

• ERROR: Unable to find the development tool cc in your path;

  yum install gcc gcc-c++ -y
  

• ERROR: Unable to find the kernel source tree for the currently running kernel

  yum install kernel-devel -y
  ./NVIDIA-Linux-x86_64-535.129.03.run --kernel-source-path=/usr/src/kernels/3.10.0-1160.105.1.el7.x86_64
  

• ERROR: Unable to load the kernel module 'nvidia.ko'

  uname -r 和 ls /usr/src/kernels，查看是否一致
  如果不一致：  
     yum update
     reboot
  ./NVIDIA-Linux-x86_64-535.129.03.run