Ubuntu由于内核升级导致NVIDIA驱动无法连接到问题

问题描述：

        服务器突然断连，重启后连接执行nvidia-smi命令显示报错“NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.”

原因：

        Ubuntu内核自动更新，导致内核与NVIDIA驱动版本不匹配。

解决方法：

1.降低内核版本，据说在开机界面选择高级配置，可以退回之前内核版本。（本人测试时无法连接服务器与显示屏，方法测试失败）

2.更新驱动

参考NVIDIA论坛问答：NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running - Linux - NVIDIA Developer Forums方法描述：

①卸载旧驱动

sudo apt-get remove --purge '^nvidia-.*'

sudo apt-get remove --purge '^libnvidia-.*'

sudo apt-get remove --purge '^cuda-.*'

②安装sudo apt-get install linux-headers-$(uname -r)

③根据自己的版本下载新驱动。CUDA Toolkit 12.2 Update 2 Downloads | NVIDIA Developer

防止再次发生：

关闭自动更新

vi /etc/apt/apt.conf.d/10periodic
vi /etc/apt/apt.conf.d/20auto-apgrades

里面的参数值都改成0。

参考

1.NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running - #5 by XuehaiPan - Linux - NVIDIA Developer Forums

2.Ubuntu由于内核升级导致NVIDIA驱动无法连接到问题 - 知乎