Detected non-NVML platform: could not load NVML: libnvidia-ml.so.1: cannot open shared object

前言

在 kubernetes 中配置 https://github.com/NVIDIA/k8s-device-plugin 时，

报错：Detected non-NVML platform: could not load NVML: libnvidia-ml.so.1: cannot open shared object

解决

kubernetes 使用运行时 docker，需要编辑通常存在的配置文件 /etc/docker/daemon.json, 以设置 nvidia-container-runtime 为默认的低级运行时：

{
    "default-runtime": "nvidia",
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}

然后重新启动 docker：

$ sudo systemctl restart docker