The fix I found to work is to add what I presume is a hardware-specific driver for mesa, namely:
# apk add mesa-dri-swrast
The driver you need may depent on the hardware – this was in a docker image on a Mac. Try the following to get a sense of what’s available:
/ # apk -U search mesa [...] mesa-gl-12.0.6-r0 mesa-dri-swrast-12.0.6-r0 mesa-gles-12.0.6-r0 mesa-dri-ati-12.0.6-r0 mesa-demos-8.3.0-r0 mesa-dev-12.0.6-r0 mesa-dri-intel-12.0.6-r0 mesa-osmesa-12.0.6-r0 mesa-gbm-12.0.6-r0 mesa-egl-12.0.6-r0 mesa-dricore-12.0.6-r0 mesa-libwayland-egl-12.0.6-r0 mesa-dri-vmwgfx-12.0.6-r0 mesa-glapi-12.0.6-r0 mesa-xatracker-12.0.6-r0 mesa-12.0.6-r0 mesa-dri-nouveau-12.0.6-r0
Now only these errors remain, although they don’t actually prevent lighthouse from succefully running:
ERROR:sandbox_linux.cc(343)] InitializeSandbox() called with multiple threads in process gpu-process. ERROR:browser_gpu_channel_host_factory.cc(125)] Failed to create channel.
Full error was:
Starting chromium, with debugger on port Waiting for 127.0.0.1:9222 ... 401:401:0318/063557.424727 :ERROR:gl_surface_glx.cc(406) glxQueryVersion failed 401:401:0318/063557.424754 :ERROR:gl_initializer_x11.cc(130) GLSurfaceGLX::InitializeOneOff failed. 401:401:0318/063557.426384 :ERROR:gpu_child_thread.cc(327)] Exiting GPU process due to errors during initialization 13:33:0318/063557.430633 :ERROR:browser_gpu_channel_host_factory.cc(113)] Failed to launch GPU process. 13:33:0318/063557.430870 :ERROR:browser_gpu_channel_host_factory.cc(113)] Failed to launch GPU process.