How-to provision an Ubuntu VM with NVIDIA vGPU

1 Overview
2 Prerequisites
3 Step by Step
4 Troubleshooting
5 Next Steps
- 5.1 To verify GPU compute with a Python ML workload:
- 5.2

Overview

This guide covers provisioning an Ubuntu 22.04 VM with NVIDIA vGPU in the VSS private cloud, installing the NVIDIA guest driver, and configuring vGPU licensing using the vss-cli. By the end, you will have a running Ubuntu VM with a licensed GPU verified via nvidia-smi.

Prerequisites

vss-cli installed. See Install the vss-cli if you haven't set it up. This guide was written against vss-cli version v2026.4.0 or later.
A VSS account with permission to provision in your group's folder.
Access to the email account that will receive the VM request notifications (one will be sent that contains the VM password).

Step by Step

Step 1 — Provision the VM

Copy thevss-cli command, fill in the variables and run it to create the VM with GPU profile:

vss-cli compute vm mk from-clib \
  --cpu 4 --gpu-profile <GPU_PROFILE> \
  --domain <FD1-or-FD2> --memory <MEM_GB> --disk <DISK_GB> \
  --os ubuntu64Guest --folder <FOLDER_NAME> \
  --source <UBUNTU_VER> \
  --description <DESC> \
  --client <CLIENT NAME> --net <NETWORK> --vss-service N/A \
  <VM_NAME>

The following example will be used for this article:

vss-cli compute vm mk from-clib \
  --cpu 4 --gpu-profile nvidia_l4-6c \
  --domain FD2 --memory 24 --disk 100 \
  --os ubuntu64Guest --folder group-v303 \
  --source ubuntu-22.04 \
  --description 'testing vss-cli vGPU deployment' \
  --client EIS --net KUALI --vss-service N/A \
  nvidia_l4-6-TEST-01

Confirm the VM was created

vss-cli compute vm ls -f name=nvidia_l4-6-TEST-01

vss-cli compute vm ls -f name=nvidia_l4-6-TEST-01
moref      name                       folder.path      cpu_count    memory_gb  power_state    ip_address
---------  -------------------------  -------------  -----------  -----------  -------------  ------------
vm-943286  2605T-nvidia_l4-6-TEST-01  VSS > Sandbox            4           24  poweredOff

Power on the VM

vss-cli --wait compute vm set vm-943286 state on

vss-cli --wait compute vm set vm-943286 state on
id                  : 62687
status              : SUBMITTED
task_id             : eecfd6cf-060f-4927-8989-aef732ba6948
message             : Request has been accepted for processing
⏳ Waiting for request 62687 to complete...
🎉 Request 62687 completed successfully:
warnings            : New state is poweredOn
errors              :

⚠️ Note: Wait for the VM to be fully provisioned and booted before proceeding (this may take several minutes)

Once the VM has powered on, check the ip address

vss-cli compute vm ls -f moref=vm-943286

vss-cli compute vm ls -f moref=vm-943286
moref      name                       folder.path      cpu_count    memory_gb  power_state    ip_address
---------  -------------------------  -------------  -----------  -----------  -------------  --------------------------------------
vm-943286  2605T-nvidia_l4-6-TEST-01  VSS > Sandbox            4           24  poweredOn      128.100.159.10 fe80::250:56ff:fe92:63c

Find the VM password in the “[VSS] - VM Request…” email

SSH to the virtual machine

ssh -p 2222 vss-admin@128.100.159.10

Step 2 — Download the NVIDIA Guest Driver

Download the NVIDIA AI Enterprise guest driver using either curl or wget:

Using curl

curl -O https://eis.utoronto.ca/ut-vss-lib/nvidia/nvidia-ai-enterprise-8.0-580.105.06-580.105.08-581.80/Guest_Drivers/nvidia-linux-grid-580_580.105.08_amd64.deb

root@nvidial4-6-test-01:~# curl -O https://eis.utoronto.ca/ut-vss-lib/nvidia/nvidia-ai-enterprise-8.0-580.105.06-580.105.08-581.80/Guest_Drivers/nvidia-linux-grid-580_580.105.08_amd64.deb
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  459M  100  459M    0     0  55.9M      0  0:00:08  0:00:08 --:--:-- 54.3M

Using wget

wget https://eis.utoronto.ca/ut-vss-lib/nvidia/nvidia-ai-enterprise-8.0-580.105.06-580.105.08-581.80/Guest_Drivers/nvidia-linux-grid-580_580.105.08_amd64.deb

vss-admin@nvidial4-6-test-01:~$ wget https://eis.utoronto.ca/ut-vss-lib/nvidia/nvidia-ai-enterprise-8.0-580.105.06-580.105.08-581.80/Guest_Drivers/nvidia-linux-grid-580_580.105.08_amd64.deb
--2026-04-29 13:19:45--  https://eis.utoronto.ca/ut-vss-lib/nvidia/nvidia-ai-enterprise-8.0-580.105.06-580.105.08-581.80/Guest_Drivers/nvidia-linux-grid-580_580.105.08_amd64.deb
Resolving eis.utoronto.ca (eis.utoronto.ca)... 128.100.102.1, 128.100.100.1
Connecting to eis.utoronto.ca (eis.utoronto.ca)|128.100.102.1|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 482112666 (460M) [application/x-debian-package]
Saving to: ‘nvidia-linux-grid-580_580.105.08_amd64.deb’

nvidia-linux-grid-580_580.105.08_amd64. 100%[===============================================================================>] 459.78M  77.7MB/s    in 6.7s

2026-04-29 13:19:52 (68.6 MB/s) - ‘nvidia-linux-grid-580_580.105.08_amd64.deb’ saved [482112666/482112666]

Step 3 — Install the NVIDIA Guest Driver

Install DKMS (Dynamic Kernel Module Support)

sudo apt install -y dkms

Install the NVIDIA driver package

sudo dpkg -i nvidia-linux-grid-580_580.105.08_amd64.deb

Step 4 — Configure vGPU Licensing

Edit the NVIDIA gridd configuration file:

sudo nano /etc/nvidia/gridd.conf

Select the FeatureType value based on your workload (I chose 4 as I plan to use the vm for an ML workload):

Value	Feature	Description

Value	Feature	Description
0	Unlicensed	Unlicensed state
1	NVIDIA vGPU	Optional, autodetected as per vGPU type
2	RTX Virtual Workstation	NVIDIA RTX Virtual Workstation (default)
4	Virtual Compute Server	Recommended for AI/ML workloads

Add or update the following line in gridd.conf:

FeatureType=4

Step 5 — Install the License Token

⚠️Note: Run Step 5 as root

Elevate to root

sudo -i

Retrieve the license token from VMware guest info and write it to the token directory

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

Set correct permissions

chmod 744 /etc/nvidia/ClientConfigToken/client_configuration_token_12-05-2023-11-26-05.tok

Exit root

exit

ℹ️ Note: The token is retrieved automatically from VMware guestinfo properties injected at provisioning time.

Step 6 — Start and Verify the License Daemon

Restart nvidia-gridd

systemctl restart nvidia-gridd

Verify GPU is visible

nvidia-smi

vss-admin@nvidial4-6-test-01:~$ nvidia-smi
Tue May  5 16:58:53 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 580.105.08             Driver Version: 580.105.08     CUDA Version: 13.0     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA L4-6C                   On  |   00000000:02:00.0 Off |                  N/A |
| N/A   N/A    P0            N/A  /  N/A  |       0MiB /   6144MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

Check license status

nvidia-smi -q | grep -i license

CleanShot 2026-04-28 at 15.48.27-20260428-194856.png

If needed you can monitor gridd logs

journalctl -u nvidia-gridd -f

Step 7 (Optional) - Release the NVIDIA license and decommission the VM

Stop the nvida-grid service

systemctl stop nvidia-gridd

vss-admin@nvidial4-6-test-01:~$ systemctl stop nvidia-gridd
==== AUTHENTICATING FOR org.freedesktop.systemd1.manage-units ===
Authentication is required to stop 'nvidia-gridd.service'.
Authenticating as: VSS User (vss-admin)
Password:
==== AUTHENTICATION COMPLETE ===

Confirm the license has been released

nvidia-smi -q | grep -i licensed

CleanShot 2026-04-29 at 10.19.10-20260429-141936.png

From your vss-cli terminal window, power off the VM

vss-cli --wait compute vm set vm-943286 state off

Decommission the vm by deleting it

vss-cli compute vm rm vm-943286

vss-cli compute vm ls -f moref=vm-943286
moref    name    folder.path    cpu_count    memory_gb    power_state    ip_address
-------  ------  -------------  -----------  -----------  -------------  ------------

Troubleshooting

Symptom	Resolution

Symptom	Resolution
`Unlicensed (Restricted)`	Compute is blocked. Check token file and restart `nvidia-gridd`.
Token signature validation failed	Token expired or driver mismatch.
`nvidia-smi` not found after install	Reboot the VM after `dpkg -i` installation.
gridd fails to start	Check `journalctl -u nvidia-gridd -f` for errors.