Multiple GPUs, for multiple VNC desktop

[LiangHuangBC]

We have one Gpu server with 4 Gpus. There are 4 developers want to use these Gpus, one for each. But we can only open one vnc desktop at the same time, when we open second, the first one will go black. Command I tried:
docker run --name glx1 -d --gpus "device=1" --privileged -it -e SIZEW=1920 -e SIZEH=1080 -e SHARED=TRUE -e VNCPASS=vncpasswd -p 5902:5901 ehfd/nvidia-glx-desktop:latest
By this command, we can run program on second gpu, but can not work together with the first container running on --gpus "device=0". The first container will go black when second container start, even they are not on same gpu.

Also tried command without privileged:
docker run --name glx1 -d --gpus "device=1" --device=/dev/tty1:rw -it -e SIZEW=1920 -e SIZEH=1080 -e SHARED=TRUE -e VNCPASS=vncpasswd -p 5902:5901 ehfd/nvidia-glx-desktop:latest
By this command, x11vnc can not start up. Error is could't open display.

And tried command in README, tried to open two vnc on same gpu 0, no luck.

And btw, there is small issue in bootstrap.sh, when docker restart your container, bootstrap will stuck on re-install driver. We fixed it by commenting it out.

[ehfd]

@LiangHuangBC I suspect the error is in privileged access so both containers can see each other. Could you post the contents of /etc/X11/Xorg.conf on both containers and also echo $NVIDIA_VISIBLE_DEVICES too? Else, you could test passing the devices HERE and disable privileged. Whether that works or not is big help to the repo because I have not tested in pure docker.

And btw, there is small issue in bootstrap.sh, when docker restart your container, bootstrap will stuck on re-install driver. We fixed it by commenting it out.

This container is only tested in Kubernetes, where when a pod restarts, the filesystem is reset. I can fix that when I have time.

Note this repo will be rebuilt when NVIDIA enables XWayland support in 470.xx drivers to use the Wayland workflow instead of Xorg.

[ehfd]

Fixed code to work in privileged mode. Please test and tell me. Not available without privileged.

bootstrap will stuck on re-install driver

Please feedback on the fix for this too.

However, ehfd/docker-nvidia-egl-desktop is guaranteed to work on multiple desktops over ehfd/docker-nvidia-glx-desktop (which uses quite hacky configurations to use Xorg inside containers) and is the recommended container to use. But ehfd/docker-nvidia-egl-desktop doesn't support Vulkan. If you need Vulkan use https://github.com/mviereck/x11docker, but you need to use docker on one single node with full root control and cannot be used on other container orchestration platforms. The ehfd/docker-nvidia-glx-desktop container is a limited functionality container designed for multi-node clusters with container orchestration or when root access is not available until Xwayland support for NVIDIA proprietary drivers are available.

[ehfd]

In progress of fixing this issue.

[ehfd]

Fixed in commit 1fbae34

Please consult the new instructions. Privileged mode is no longer supported in favor of unprivileged containers.

[ehfd]

Added in Documentation.