I’m testing CasparCG Server 2.5.1 on Ubuntu 24.04. So far, I’ve only been playing video files, but during testing, the server unexpectedly exits without recording any errors in the CasparCG log.
However, I captured the following messages on Linux:
ERROR:content/browser/gpu/gpu_process_host.cc:999 GPU process exited unexpectedly: exit_code=139
WARNING:content/browser/gpu/gpu_process_host.cc:1441 The GPU process has crashed 3 time(s)
WARNING:ui/gfx/linux/gpu_memory_buffer_support_x11.cc:49 dri3 extension not supported.
ERROR:ui/gl/angle_platform_impl.cc:42 DisplayVkXcb.cpp:62 (initialize): xcb_connect() failed, error 1
ERROR:ui/gl/gl_display.cc:639 eglInitialize SwANGLE failed with error EGL_NOT_INITIALIZED
ERROR:components/viz/service/main/viz_main_impl.cc:189 Exiting GPU process due to errors during initialization
The system log also contains this entry:
out 08 15:11:34 play3 kernel: traps: MemoryInfra[2930] trap invalid opcode ip:78b8a1b270aa sp:78b83affb730 error:0 in libcef.so
I need to get this machine ready for production. It will also be used to play HTML templates, so disabling CEF is not a viable solution for my setup.
Has anyone encountered a similar issue with CasparCG 2.5.1 on Ubuntu 24.04? Are there any recommended installation steps, dependencies, GPU driver settings, or configuration changes that could help resolve these crashes?
I read through that thread, but the issue described there occurs when using an HTML template. My issue is different: it occurs during video playback, after repeatedly playing different videos in a loop for some time.
You sure that no html content had been started? Because those errors and logs are without a doubt from the html module. But I guess it’s possible that it can fail after just being initialised, without playing anything. I would suggest trying the suggested solution on that thread anyway
Some of these CEF HTML related issues have been patched since 2.5.1 in the master branch. If we can isolate which errors still exist and what environment produces them we can work on patching the rest.
I can confirm from testing development builds that the crash is fixed on master, but some of the GPU errors from CEF persist from default settings. Interestingly enough since I am running casparcg inside a slot in my new frontend program those errors go to the console output of the parent process of the frontend instead of showing up in the casparcg console output so this is definately something that needs debugged in CEF but could still relate to something in how we are using it.