NVIDIA Kubernetes installation Pre-requisites

Graphics / Linux Linux
1

Hello Team,

I am New to the NVIDIA World. What i want to do new is i want to install a NVIDIA Kubernetes in one of our Physical servers. Can you please help me with the installation steps in short and also the pre-requisites of the same.

Please note: i have already installed the Docker version 18.06 and i also have a kubernetes version 1.11.3.

Please let me know what i need to Do if i need to have the NVIDIA GPU feature for my server.

Thank you
Naveen Kulkarni

2

[url]https://docs.nvidia.com/datacenter/kubernetes-install-guide/index.html[/url]
The installation of the driver depends on your distribution, look for a repository.

3

Hello Generix,

Thank you for replying. I went through the document completely. I i followed the same steps. But i am unable to run my app with the nividia kubernetes support it give me the below error. Can you please help me on this.

raise ImportError("""Please make sure you have pycuda installed""")

ImportError: Please make sure you have pycuda installed

I also tried to run for another app with nvidia kubernetes and i get this below error.

AssertionError:
Found no NVIDIA driver on your system. Please check that you
have an NVIDIA GPU and installed a driver from

4

I presume you’re using Ubuntu as the base OS, did you already install the driver/cuda? Install method:

  • Don’t use the .run installers, use --uninstall to uninstall them
  • purge anthing nvidia/cuda
  • add the ubuntu graphics ppa
  • install the driver from that
  • download the cuda .deb
  • add it to your system
  • don’t install cuda
  • instead, run sudo apt install cuda-toolkit-10-0

After that, you can follow this:
[url]https://thenewstack.io/build-a-machine-learning-testbed-based-on-kubernetes-and-nvidia-gpu/[/url]

Addendum: cuda 10 requires nvidia docker v2

5

Hello Generix,

Thank you for replying, Can you please help me with the minimum documentation and minimum steps to configure this. Please let me know where i can find everything in a single places. Starting from the CUDA to NVIDIA kubernetes.

Please note: Currently i have the docker version 18.06 installed in my machine. I also have the Kubernetes version 1.11.3. I can run our normal Algo’s, I only need the nividia support to run few Algo’s of mine which strictly runs only on the nvidia runtime. Henceforth, i want to install NVIDIA Kubernetes.

Thanks and Regards,
Naveen Kulkarni