How-to configure and licence a provisioned vGPU

When you provision a vGPU on The ITS Private Cloud, a couple of steps must be performed before using the resource. First, a driver must be installed in the guest operating system and a license must be assigned. This how-to provides the necessary steps to get your vGPU ready for action.

Download and Install the driver

1. From the Guest Operating system open a sftp, scp or your vskey-stor.eis.utoronto.ca MinIO interface and browse to the following path /ut-vss-lib/nvidia-grid-vpshere-7.0-525.147.01-525.147.05-529.19/Guest_Drivers.

2. Download the driver based on your operating system. i.e. using sftp to install nvidia-linux-grid on Ubuntu:

   sftp {vss-user}@vskey-stor.eis.utoronto.ca cd /ut-vss-lib/nvidia-grid-vsphere-7.0-535.183.04-535.183.01-538.67/Guest_Drivers get nvidia-linux-grid-535_535.183.01_amd64.deb

3. Install the driver in the virtual machine guest operation system. i.e. on Ubuntu:

   apt install dkms dpkg -i nvidia-linux-grid-535_535.183.01_amd64.deb

License your vGPU

Windows

1. Open cmd as Administrator.

2. cd into C:\Program Files\VMware\VMware Tools

   cd C:\Program Files\VMware\VMware Tools

3. Extract the token from the guest info interface:

   .\vmtoolsd.exe --cmd "info-get guestinfo.ut.vss.nvidia_token" > "C:\Program Files\NVIDIA Corporation\vGPU Licensing\ClientConfigToken\client_configuration_token_12-05-2023-11-26-05.tok"

4. Restart services:

   net stop "NVDisplay.ContainerLocalSystem" net start "NVDisplay.ContainerLocalSystem"

5. The virtual machine will check-in with our on-premise license server and activate any requested feature.

Linux

1. As root, copy the file /etc/nvidia/gridd.conf.template to /etc/nvidia/gridd.conf.

2. Open a text editor and edit the FeatureType to select either of the following values:

   # Description: Set Feature to be enabled # Data type: integer # Possible values: # 0 => for unlicensed state # 1 => for NVIDIA vGPU # 2 => for NVIDIA RTX Virtual Workstation # 4 => for NVIDIA Virtual Compute Server

3. To enable NVIDIA RTX Virtual Workstation, select 2:

   FeatureType=2

4. Extract the token from the guest info interface and save it to a file in the directory /etc/nvidia/ClientConfigToken/ as follows:

   echo -n -e $(vmware-rpctool "info-get guestinfo.ut.vss.nvidia_token") > /etc/nvidia/ClientConfigToken/client_configuration_token_12-05-2023-11-26-05.tok

5. If necessary, change the mode of the client configuration token to 744:

   chmod 744 /etc/nvidia/ClientConfigToken/client_configuration_token_*.tok

6. Restart the services.

   systemctl restart nvidia-gridd

7. The virtual machine will check-in with our on-premise license server and activate any requested feature.

References

• Client Licensing User Guide