OpenGL acceleration with Docker

[conroy-cheers]

Has anyone managed to get an OpenGL accelerated Docker container running behind WSLG?
I can get GUI apps (e.g. Gazebo) running great with acceleration in a WSL2 Ubuntu 20.04 distro, just by upgrading Mesa. However, I haven't figured out how to do this in a container. My docker-compose config maps the following volumes and environment variables, and Gazebo runs in the container, but performance is poor, as only llvmpipe software rendering is being used.

  volumes:
    - /tmp/.X11-unix:/tmp/.X11-unix 
    - /mnt/wslg:/mnt/wslg
    - /usr/lib/wsl:/usr/lib/wsl
    - /dev/dxg:/dev/dxg
  environment:
    - DISPLAY
    - WAYLAND_DISPLAY
    - XDG_RUNTIME_DIR
    - PULSE_SERVER

I've tried this config with both Docker Desktop (which runs Docker in its own WSL2 distro), and Docker installed directly in my WSL2 Ubuntu 20.04 distro, but the results are identical either way.
What's the magic sauce? How does Mesa detect the d3d12 backend? According to this blog post, the userspace just needs Mesa >=21.0, libd3d12/libdxcore, and /dev/dxg present. libd3d12.so and libdxcore.so are both mapped through /usr/lib/wsl, /dev/dxg is mapped, and Mesa 21.1.5 is installed in the container.

Any help would be greatly appreciated :)

[onomatopellan]

For GPU support inside the container from the command line you need to run docker with the argument --gpus=all.
For docker-compose you need to add the gpu capabilities.

[conroy-cheers]

This works for CUDA, but doesn't seem to help for OpenGL

[onomatopellan]

I see. What's the output for glxinfo -B inside the container?

[dwbertol]

I'm facing the same problem trying to overcome it without success for a while.
My glxinfo -B inside docker is

name of display: :0
display: :0  screen: 0
direct rendering: Yes
Extended renderer info (GLX_MESA_query_renderer):
    Vendor: Mesa/X.org (0xffffffff)
    Device: llvmpipe (LLVM 12.0.0, 256 bits) (0xffffffff)
    Version: 21.0.3
    Accelerated: no
    Video memory: 11898MB
    Unified memory: no
    Preferred profile: core (0x1)
    Max core profile version: 4.5
    Max compat profile version: 3.1
    Max GLES1 profile version: 1.1
    Max GLES[23] profile version: 3.2
OpenGL vendor string: Mesa/X.org
OpenGL renderer string: llvmpipe (LLVM 12.0.0, 256 bits)
OpenGL core profile version string: 4.5 (Core Profile) Mesa 21.0.3
OpenGL core profile shading language version string: 4.50
OpenGL core profile context flags: (none)
OpenGL core profile profile mask: core profile

OpenGL version string: 3.1 Mesa 21.0.3
OpenGL shading language version string: 1.40
OpenGL context flags: (none)

OpenGL ES profile version string: OpenGL ES 3.2 Mesa 21.0.3
OpenGL ES profile shading language version string: OpenGL ES GLSL ES 3.20

[onomatopellan]

The team just posted a sample of a container with accelerated OpenGL.

https://github.com/microsoft/wslg/tree/main/samples/container

After looking at that this worked for me:

docker run -it --rm -v /tmp/.X11-unix:/tmp/.X11-unix -v /mnt/wslg:/mnt/wslg -v /usr/lib/wsl:/usr/lib/wsl --device=/dev/dxg -e DISPLAY=$DISPLAY -e WAYLAND_DISPLAY=$WAYLAND_DISPLAY -e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR -e PULSE_SERVER=$PULSE_SERVER -e LD_LIBRARY_PATH=/usr/lib/wsl/lib --gpus all ubuntu/latest:opengl2

I guess the most important part was using the env variable LD_LIBRARY_PATH=/usr/lib/wsl/lib