NVIDIA Driver Not Available After Reprovisioning GPU Nodes Using GPU OS Image

[shivshankar7977]

Hi All,

I have a setup with 2 GPU compute nodes and 1 controller node.

The NVIDIA GPU driver was installed manually on both GPU nodes, one by one. After successful installation, an OS image (GPU image) was captured from one of the GPU nodes. using luna node osgrab command.

However, when the GPU nodes were reprovisioned using this GPU OS image, the NVIDIA driver was not available on the nodes (the driver was not showing / nvidia-smi not working).

Thanks!

[bartlamboo]

Hi @shivshankar7977 ,

Just out of curiosity, was the kernel upgraded when doing the driver install as well?
Because if so, it would have built the kernel modules for the incorrect kernel version. (kernel versions are locked)
Also, did you make sure to run the --nodry version of the osgrab?