Needed a Guide to setup Devpod , on two ubuntu servers . where one server having NAS as storage to export NFS and another one having A100 GPU .

[sanjay7178]

So with the setup , I want to make kubernetes cluster on both servers and want to give access to user to use GPU , CPU , Storage and Memory

[pratikrath126]

Hey! If you want DevPod across two Ubuntu servers (one with NAS storage, one with A100 GPU), I’d approach it like this:

1. Build the K8s cluster first (e.g., kubeadm or k3s) with both nodes joined.
2. GPU node: install NVIDIA drivers + container runtime (nvidia-container-toolkit) and deploy the NVIDIA device plugin so pods can request GPU resources.
3. NAS node: export NFS from the NAS host and set up the NFS CSI driver in the cluster. Then you can create PersistentVolumes for shared storage.
4. DevPod: configure DevPod to target your Kubernetes cluster. DevPod uses devcontainer images, so define CPU/GPU resource requests in your devcontainer config or the DevPod template.

Once those pieces are in place, users can select GPU-enabled workspaces and shared storage via PVCs. The DevPod docs have a Kubernetes provider setup guide that should map directly to this topology.

If you want, I can help sketch a minimal kubeadm + NFS + GPU plugin checklist.

[sanjay7178]

Hey! If you want DevPod across two Ubuntu servers (one with NAS storage, one with A100 GPU), I’d approach it like this:

1. Build the K8s cluster first (e.g., kubeadm or k3s) with both nodes joined.
2. GPU node: install NVIDIA drivers + container runtime (nvidia-container-toolkit) and deploy the NVIDIA device plugin so pods can request GPU resources.
3. NAS node: export NFS from the NAS host and set up the NFS CSI driver in the cluster. Then you can create PersistentVolumes for shared storage.
4. DevPod: configure DevPod to target your Kubernetes cluster. DevPod uses devcontainer images, so define CPU/GPU resource requests in your devcontainer config or the DevPod template.

Once those pieces are in place, users can select GPU-enabled workspaces and shared storage via PVCs. The DevPod docs have a Kubernetes provider setup guide that should map directly to this topology.

If you want, I can help sketch a minimal kubeadm + NFS + GPU plugin checklist.

Thanks 👍