离线环境安装 NVIDIA Container Toolkit

  1. 环境准备
  2. 包文件安装

环境准备

  • 已正确安装对应GPU驱动
  • NVIDIA drivers ~= 384.81
  • nvidia-docker >= 2.0 || nvidia-container-toolkit >= 1.7.0
  • nvidia-container-runtime configured as the default low-level runtime
  • Kubernetes version >= 1.10

包文件安装

  • 根据对应系统环境下载离线安装包

  • 各包文件依赖关系

  • 安装顺序

    • ibnvidia-container
    • libnvidia-container-tools
    • nvidia-container-toolkit
    • nvidia-container-runtime
  • 配置docker环境则需要安装nvidia-docker2并配置nvidia-container-runtime

    vim /etc/docker/daemon.json
    {"default-runtime":"nvidia","runtimes": {"nvidia": {"path":"/usr/bin/nvidia-container-runtime","runtimeArgs": []
            }
        }
    }
    systemctl restart docker
    
  • 配置containerd环境

    vim /etc/containerd/config.toml
    version = 2
    [plugins."io.containerd.grpc.v1.cri".containerd]
      default_runtime_name = "nvidia"
      [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia]
        privileged_without_host_devices = false
        runtime_engine = ""
        runtime_root = ""
        runtime_type = "io.containerd.runc.v2"
        [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.nvidia.options]
          BinaryName = "/usr/bin/nvidia-container-runtime"
    systemctl restart containerd
    

转载请注明来源,欢迎对文章中的引用来源进行考证,欢迎指出任何有错误或不够清晰的表达。可以在下面评论区评论,也可以邮件至 365433079@qq.com