libGL error: failed to load driver swrast in docker container

Question

I have a docker container from which I am trying to run a pyqt app. Everything works well except a chunk of the GUI is not able to render. The docker logs throw this out:

libGL error: failed to load driver: swrast
X Error: GLXBadContext 169
 Extension:    154 (Uknown extension)
 Minor opcode: 6 (Unknown request)
 Resource id:  0x6400003
X Error: BadValue (integer parameter out of range for operation) 2
 Extension:    154 (Uknown extension)
 Minor opcode: 3 (Unknown request)
 Resource id:  0x0
...
QGLContext::makeCurrent(): Failed.

In my Dockerfile, I tried installing pretty much all the packages I could find that might be related, including mesa-utils.

In terms of the docker-compose file, here's what it looks like:

version: '2'
    services:
    gui:
        build: .
        volumes:
        - .:/usr/src
        - /tmp/.X11-unix:/tmp/.X11-unix
        command: /bin/bash -c "python start.py"
        environment:
        - DISPLAY=unix$DISPLAY
        - QT_X11_NO_MITSHM=1
        devices:
        - "/dev/snd:/dev/snd"
        - "/dev/dri:/dev/dri"
        privileged: true

Any ideas what I might be missing?

Answer 1

Figured it out. I had to build the gui with hardware accelerated OpenGL support. Theres a repo (https://github.com/gklingler/docker3d) that contains docker images with nvidia or other graphics drivers support.

The other catch was, it didn't work for me unless the host and the container had the exact same driver. In order to resolve this, you can run the following shell script if you're running on linux:

#!/bin/bash
version="$(glxinfo | grep "OpenGL version string" | rev | cut -d" " -f1 | rev)"
wget http://us.download.nvidia.com/XFree86/Linux-x86_64/"$version"/NVIDIA-Linux-x86_64-"$version".run
mv NVIDIA-Linux-x86_64-"$version".run NVIDIA-DRIVER.run