I have packed the project, tested ok in Windows., cross-compile linux,
however, when deploying in a ubuntu docker,
I encounter following error (when running with xxx_project.sh -RenderOffscreen -PixelStreamingIP=localhost -PixelStreamingPort=8888) :
[2022.07.12-14.25.08:410][ 0]LogInit: Display: Game Engine Initialized.
[2022.07.12-14.25.08:410][ 0]LogCUDA: Display: Initialising CUDA API...
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: CUDA API initialised successfully.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found 7 CUDA capable devices.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 0 called NVIDIA A100-PCIE-40GB.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 1 called NVIDIA Quadro RTX 8000.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 2 called NVIDIA Quadro RTX 8000.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 3 called NVIDIA Quadro RTX 8000.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 4 called NVIDIA Quadro RTX 8000.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 5 called NVIDIA Quadro RTX 8000.
[2022.07.12-14.25.08:509][ 0]LogCUDA: Display: Found device 6 called NVIDIA Quadro RTX 8000.
Failed to find symbol file, expected location:
"/home/hmp/fuqingxu/UHMP/Build/LinuxNoEditor/UHMP/Binaries/Linux/UHMP.sym"
Fatal error: [File:F:/UnrealSourceCode/UnrealEngine-4.27.2-release/Engine/Source/Runtime/CUDA/Source/Private/CudaModule.cpp] [Line: 202]
CUDA module failed to create a CUDA context on the RHI selected device with UUID 2010431984.
[2022.07.12-14.25.08:512][ 0]LogCore: Error: appError called: Fatal error: [File:F:/UnrealSourceCode/UnrealEngine-4.27.2-release/Engine/Source/Runtime/CUDA/Source/Private/CudaModule.cpp] [Line: 202]
CUDA module failed to create a CUDA context on the RHI selected device with UUID 2010431984.
0x0000000008bc370b UHMP!UnknownFunction(0x89c370a)
0x0000000008bc5dbf UHMP!UnknownFunction(0x89c5dbe)
0x0000000003362b9c UHMP!UnknownFunction(0x3162b9b)
0x0000000003354063 UHMP!UnknownFunction(0x3154062)
0x00000000033629c0 UHMP!UnknownFunction(0x31629bf)
0x0000000008b56d69 UHMP!UnknownFunction(0x8956d68)
0x00007f17d57d0083 libc.so.6!__libc_start_main(+0xf2)
0x0000000003353649 UHMP!UnknownFunction(0x3153648)
Signal 11 caught.
Malloc Size=65538 LargeMemoryPoolOffset=65554
CommonUnixCrashHandler: Signal=11
[2022.07.12-14.25.08:524][ 0]LogCore: === Critical error: ===
Unhandled Exception: SIGSEGV: invalid attempt to write memory at address 0x0000000000000003
[2022.07.12-14.25.08:524][ 0]LogCore: Fatal error: [File:F:/UnrealSourceCode/UnrealEngine-4.27.2-release/Engine/Source/Runtime/CUDA/Source/Private/CudaModule.cpp] [Line: 202]
CUDA module failed to create a CUDA context on the RHI selected device with UUID 2010431984.
please help, I don’t know what to do with this crash …
detail___
the docker contrainer is launched by
docker run -itd --name uebuild-$USER \
--runtime=nvidia \
-e NVIDIA_DRIVER_CAPABILITIES=all \
--net host \
--memory 400G \
--memory-reservation 400G \
--volume /var/run/docker.sock:/var/run/docker.sock \
--gpus all \
--shm-size=16G \
my_nv_docker_image
And the game is launched by :
my_project.sh -RenderOffscreen -nosound -PixelStreamingIP=localhost -PixelStreamingPort=8888
Whole startup log:
UHMP.log (62.5 KB)
end of detail____