docker: Error response from daemon: could not select device driver “” with capabilities: [[gpu]].

该问题是**缺少NVIDIA Container Toolkit**导致的
服务器若有网络，则

# 添加GPG密钥和仓库
distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
curl -s -L https://nvidia.github.io/libnvidia-container/$distribution/libnvidia-container.list | \
    sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
    sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

# 更新并安装
sudo apt-get update
sudo apt-get install -y nvidia-container-toolkit

# 配置Docker daemon
sudo nvidia-ctk runtime configure --runtime=docker

# 重启Docker服务
sudo systemctl restart docker

若服务器离线，则

1. 先在有网的服务器上下载相关包
# 创建下载目录
mkdir nvidia-container-toolkit-offline
cd nvidia-container-toolkit-offline

# 下载包及其依赖
# Ubuntu 22.04 所需的包（版本1.14.3）
wget https://nvidia.github.io/libnvidia-container/stable/ubuntu22.04/amd64/libnvidia-container1_1.14.3-1_amd64.deb

wget https://nvidia.github.io/libnvidia-container/stable/ubuntu22.04/amd64/libnvidia-container-tools_1.14.3-1_amd64.deb

wget https://nvidia.github.io/nvidia-container-runtime/stable/ubuntu22.04/amd64/nvidia-container-toolkit-base_1.14.3-1_amd64.deb

wget https://nvidia.github.io/nvidia-container-runtime/stable/ubuntu22.04/amd64/nvidia-container-toolkit_1.14.3-1_amd64.deb

# 打包
tar -czf nvidia-container-toolkit.tar.gz *.deb

2. 在离线服务器上操作
# 解压
tar -xzf nvidia-container-toolkit.tar.gz

# 安装所有deb包
sudo dpkg -i *.deb

# 如果有依赖问题，尝试修复
sudo apt-get install -f

# 配置Docker daemon
sudo nvidia-ctk runtime configure --runtime=docker

# 查看配置是否生成
cat /etc/docker/daemon.json

# 重启Docker
sudo systemctl restart docker

# 检查Docker状态
sudo systemctl status docker

# 重启服务

{
    "runtimes": {
        "nvidia": {
            "path": "nvidia-container-runtime",
            "runtimeArgs": []
        }
    }
}