Error response from daemon: unknown or invalid runtime name: nvidia

TrueNAS General
1

Hi everyone,

I think I’ve stumbled across an extremely frustrating error in the latest version of TrueNAS. I recently removed my NVIDIA GPU from my system, but failed to uninstall the drivers before doing so. Because of this, I’m unable to start any Docker containers and I’m also unable to uninstall the drivers. All of the options necessary to manage the GPU drivers are not visible in the UI.

Docker / Dockge throws this error when starting up:
Error response from daemon: unknown or invalid runtime name: nvidia

Great. Now I have to drive back to the datacenter, un-rack the server, install the GPU, remove the drivers, and button everything back up. Ugh.

Cheers.

1 Like
2

I figured out how to disable NVIDIA support in Docker and remove the driver from TrueNAS in case anyone runs into this in the future.

# Check whether the NVIDIA drivers are loaded
systemd-sysext

# HIERARCHY EXTENSIONS SINCE                      
# /opt      none       -                          
# /usr      nvidia     Thu 2025-08-14 19:06:02 PDT

# Remove the NVIDIA overlay:
sudo systemd-sysext unmerge

# Check where the image is located
systemd-sysext list

# NAME   TYPE PATH                          TIME
# nvidia raw  /run/extensions/nvidia.raw    ...

# Install dev tools to enable write access to the overlay
sudo install-dev-tools

# Remove the image
sudo rm /usr/share/truenas/sysext-extensions/nvidia.raw

# Disable the NVIDIA driver in the docker config (remove the flag that is currently invisible in the UI)
sudo midclt call --job docker.update '{"nvidia": false}'

# Restart Docker to apply
sudo systemctl restart docker

# Verify that the change stuck
sudo midclt call docker.config

# Reboot the server to complete the changes