Getting NVidia GPUs to work in your kubernetes cluster has been in my experience a hot mess with conflicting documentation and helm charts. Through the power of chutzpah and dumb luck I managed to get a tutorial made to provide a working setup.
For this tutorial I assume you have the following:
- A Working K3S cluster with at least one node with some sort of NVidia GPU
- Helm installed
This guide has been tested on:
- Debian 13
- Ubuntu 24.04
Instructions for Talos Linux can be found here: https://docs.siderolabs.com/talos/v1.12/configure-your-talos-cluster/hardware-and-drivers/nvidia-gpu-proprietary Do note that you need to install Node Feature Discovery for the NVidia helm chart to work properly
The first step is to make sure the OS itself can interface with your GPU:
If you installed Debian normally you would want to edit /etc/apt/sources.list, the first line is likely the correct one to edit:
sudo nano /etc/apt/sources.list
deb http://mirror.csclub.uwaterloo.ca/debian/ trixie main
Stick contrib non-free non-free-firmware onto the end so that it looks something like this:
deb http://mirror.csclub.uwaterloo.ca/debian/ trixie main non-free-firmware contrib
Press Ctrl+X to save, then Y to confirm and then enter.
sudo sed -i 's/^Components: main$/Components: main contrib non-free non-free-firmware/' /etc/apt/sources.list.d/debian.sources
sudo apt-get update
sudo apt-get install nvidia-driver nvidia-smi
You will get a warning about the nouveau driver, the open source driver for nvidia gpus. Just press okay.
Installing the NVidia drivers on Ubuntu is very easy:
sudo apt install nvidia-dkms-590-server nvidia-utils-590-server
Now we'll install gpg and add the nvidia container toolkit repository:
sudo apt-get install -y gpg curl
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg
curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list
sudo apt-get update
sudo apt-get install firmware-misc-nonfree nvidia-container-runtime
And now we reboot:
sudo reboot
If you want to make sure everything is working, log back in and run nvidia-smi. This is an optional utility to see what's going on with your GPU(s):
nvidia-smi
Fri Jan 30 01:41:02 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.163.01 Driver Version: 550.163.01 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| 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 Quadro P400 Off | 00000000:06:10.0 Off | N/A |
| 40% 53C P8 N/A / N/A | 2MiB / 2048MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+
No more needs to be done directly on your Kubernetes nodes.
Node Feature Discovery adds labels to your nodes based on what hardware is available and what features your CPU(s) support. The NVidia Device Plugin needs these labels for scheduling.
helm repo add node-feature-discovery https://kubernetes-sigs.github.io/node-feature-discovery/charts
helm repo update
helm upgrade --install node-feature-discovery node-feature-discovery/node-feature-discovery --create-namespace --namespace node-feature-discovery
Next we'll install the NVidia K8s Device Plugin using helm. Helm is like apt-get but for Kubernetes.
helm repo add nvdp https://nvidia.github.io/k8s-device-plugin
helm repo update
By default if you have only one GPU installed only one pod can use that resource. In values.yaml, line 18 tells the device plugin to treat that one GPU as if it were actually four GPUs.
And now we will install the K8s Device Plugin:
helm upgrade --install nvdp nvdp/nvidia-device-plugin --create-namespace --namespace nvidia --values values.yaml
I highly recommend using K9S to see if everything's going well. The whole setup process will take a minute or two to start up. If everything has gone well all the pods should be Running:
kubectl get pods -n nvidia
NAME READY STATUS RESTARTS AGE
nvdp-node-feature-discovery-gc-5f949b6d-zhkg4 1/1 Running 0 18m
nvdp-node-feature-discovery-master-b85494b56-qhwx4 1/1 Running 0 18m
nvdp-node-feature-discovery-worker-72fck 1/1 Running 0 18m
nvdp-nvidia-device-plugin-cbfz4 2/2 Running 0 18m
nvdp-nvidia-device-plugin-gpu-feature-discovery-t9dxw 2/2 Running 0 18m
Let's see if it works:
kubectl apply -f nvidia-smi.yaml
Let's see if it worked:
kubectl logs nvidia-smi
Sun Mar 8 02:20:49 2026
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.163.01 Driver Version: 550.163.01 CUDA Version: 12.5 |
|-----------------------------------------+------------------------+----------------------+
| 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 Quadro P400 On | 00000000:01:00.0 Off | N/A |
| 46% 59C P8 N/A / N/A | 2MiB / 2048MiB | 0% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| No running processes found |
+-----------------------------------------------------------------------------------------+
It worked! You can remove the test:
kubectl delete -f nvidia-smi.yaml
Getting other containers to use NVidia GPUs is fairly easy. Look at the contents of jellyfin.yaml, specifically lines 23 and 27-29. Line 23 tells Kubernetes to use the nvidia runtime instead of the default containerd and lines 27 to 29 set GPU limits. The nvidia runtime will automatically copy everything needed for your pod to use the GPU.
- https://docs.siderolabs.com/talos/v1.12/configure-your-talos-cluster/hardware-and-drivers/nvidia-gpu-proprietary
- https://github.com/mirceanton/home-ops/blob/6c8f7e23d7531489eaa273500fa37fbb2ea523a2/apps/kube-system/nvidia-device-plugin/app/helm-release.yaml
If you want to easily spin up virtual machines that auto-install the NVidia driver and toolkit checkout my Ubuntu-CloudInit-Docs repository.
Check out my guide for using Intel GPUs in Kubernetes: https://github.com/UntouchedWagons/K3S-Intel
