[BUG] GPU issues on Unraid
Ninguém assumiu esta issue ainda.
Avaliação
- Dificuldade
- 4/5
- Tempo estimado
- 3-5 dias
- Facilidade para iniciantes
- 35/100
- Tipo de issue
- Bug
- Clareza
- Precisa de esclarecimento
- Status de atividade
- Ativa
- Stack de tecnologia
- docker, docker-compose
- Domínio
- computer-graphics, desktop
Direção de pesquisa
Start by reproducing the supplied Docker Compose configuration and reviewing the container logs around /dev/dri, NVIDIA devices, and the KMS error. Trace why Wayland cannot allocate a GBM buffer; done when the container initializes the GPU renderer instead of falling back to software rendering.
Escrita pelo modelo de indexação a partir do texto da issue.
Descrição
Is there an existing issue for this?
- I have searched the existing issues
Current Behavior
GPUs cannot be enabled on Unraid. This was raised in a issue previously, but the thread seemed to be abandoned with no solution.
Expected Behavior
It should use the GPU...
Steps To Reproduce
Container boots but fails to get permission for the GPU and falls back to software rendering. The GPU(s) are detected (a 3080 on D128 and a Tesla P40 on D129), everything seems to go smoothly until
KMS: DRM_IOCTL_MODE_CREATE_DUMB failed: Permission denied
[Wayland] GPU renderer failed to initialize (Failed to allocate GBM buffer); rendering in software (Pixman).
I have tried every combination of docker configs I can think of with no luck.
- Passing /dev/dri
- Defining DINODE and DRI_NODE
- modeset is configured in the bootloader and a
cat /sys/module/nvidia_drm/parameters/modesetreturns Y - Adding NVIDIA_VISIBLE_DEVICES and NVIDIDA_DRIVER_CAPABILITIES (as many Unraid docker containers use)
- Running
nvidia-modprobe --modesetafter reboot and before launching the container
Environment
- OS: Unraid 7.1.4
- How docker service was installed: Compose
- Nvidia Driver: 580.82.09
- GPU: RTX 3080 (Tesla P40 installed but not trying to use that)
CPU architecture
x86-64
Docker creation
services:
webstation:
image: lscr.io/linuxserver/webstation:latest
container_name: webstation
runtime: nvidia
gpus: all
environment:
- PUID=99
- PGID=100
- TZ=America/Los_Angeles # Change to your timezone
- DRINODE=/dev/dri/renderD128
- DRI_NODE=/dev/dri/renderD128
devices:
- /dev/dri:/dev/dri
- /dev/nvidia-modeset:/dev/nvidia-modeset
volumes:
- /mnt/user/appdata/webstation:/config # Replace with your local directory
- /mnt/user/Media/ROMs/roms:/romm-roms
- /mnt/user/Media/ROMs/BIOS:/bios
ports:
- 3000:3000
- 3001:3001
shm_size: "1gb"
restart: unless-stopped
Container logs
User UID: 99
User GID: 100
───────────────────────────────────────
[ls.io-init] Generating labwc rc.xml from template
[ls.io-init] Disabling auto-maximization
/usr/bin/nvidia-smi
**** adding /dev/dri/renderD129 to video group root with id 0 ****
**** permissions for /dev/dri/renderD128 are good ****
**** permissions for /dev/dri/card2 are good ****
**** permissions for /dev/dri/card1 are good ****
**** creating video group video970t with id 18 ****
**** adding /dev/dri/card0 to video group video970t with id 18 ****
by-path
card0
card1
card2
renderD128
renderD129
**** NVIDIA GPU detected ****
**** Setting up OpenCL ICD for NVIDIA ****
**** Setting up EGL vendor file for NVIDIA ****
**** Routing GLX to the NVIDIA library ****
[custom-init] No custom files found, skipping...
[svc-de] Wayland mode: Waiting for socket at /config/.XDG/wayland-1...
[ls.io-init] done.
19
20
[pixelflux] Render node 0 encodes H264, H265 on nvenc.
[pixelflux] Thread priority -15 refused and rtkit does not grant it; capture threads run at normal priority.
[Wayland] Renderer: GL on /dev/dri/renderD128.
pci id for fd 43: 10de:2206, driver (null)
pci id for fd 44: 10de:2206, driver (null)
pci id for fd 45: 10de:2206, driver (null)
pci id for fd 44: 10de:2206, driver (null)
pci id for fd 45: 10de:2206, driver (null)
pci id for fd 42: 10de:2206, driver (null)
pci id for fd 46: 10de:2206, driver (null)
pci id for fd 47: 10de:2206, driver (null)
pci id for fd 46: 10de:2206, driver (null)
pci id for fd 47: 10de:2206, driver (null)
KMS: DRM_IOCTL_MODE_CREATE_DUMB failed: Permission denied
[Wayland] GPU renderer failed to initialize (Failed to allocate GBM buffer); rendering in software (Pixman).
[Wayland] View 1 created over output 0: 1024x768 @ (0, 0).
[Wayland] Socket listening on: "wayland-1"
INFO:server:Wayland compositor socket: wayland-1
INFO:main:Selkies 2.0.0 starting: websockets transport, Wayland capture, encoder h264enc at 8-240 fps, sized by the client, audio on, gamepads on, access op
en.
INFO:server:Starting service: websockets
INFO:gamepad:4 virtual gamepads ready ('Microsoft X-Box 360 pad', sockets under /tmp); kernel uinput off.
INFO:input:Wayland clipboard: native compositor callback active (no polling).
- Linguagem predominante
- Dockerfile
- Estrelas
- 18
- Forks
- 5
- Métricas de merge de PRs
- Nenhum PR com merge em 30d
Preparar o ambiente
- Inclui um Dockerfile ou arquivo Docker Compose
- Tem um modelo de pull request
- Ler o guia de contribuição
Primeiros passos
- Leia a issue inteira e depois o guia de contribuição do projeto.
- Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
- Faça um fork do repositório e trabalhe em uma branch.
- Abra um pull request que referencie o número da issue.
Issues semelhantes
-
area: engine enhancement
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 75/100
Mantenedores costumam responder em até 1 dia
-
bug
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 62/100
MrNeRF/LichtFeld-Studio#4107 ·
Mantenedores costumam responder em até 1 dia
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 76/100
Mantenedores costumam responder em até 1 dia
-
Dificuldade 2/5 1-3 horas Facilidade para iniciantes 64/100
Mantenedores costumam responder em até 1 dia
-
bug good first issue P3 svg
Dificuldade 1/5 Menos de uma hora Facilidade para iniciantes 87/100
Mantenedores costumam responder em até 1 dia