Server Version#: 1.43.0.10492
Player Version#:
<If providing server logs please do NOT turn on verbose logging, only debug logging should be enabled>
I am running plex in docker on Debian 12.
I have a Nvidia Quadro P600 GPU.
I have a valid plex pass.
container I am using is the lcd.io/linuxserver/plex:latest
I have installed the Nvidia drivers and container toolkit. Nvidia driver version is 550.163.01
In my docker compose file, I have the following added:
Under environment:
- NVIDIA_VISIBLE_DEVICES=0
- NVIDIA_DRIVER_CAPABILITIES=all
Under Devices:
# CDI GPU injection (must be first)
**-** nvidia.com/gpu=all
*# Optional: keep explicit device mappings*
**-** /dev/dri:/dev/dri
**-** /dev/nvidia0:/dev/nvidia0
**-** /dev/nvidiactl:/dev/nvidiactl
**-** /dev/nvidia-uvm:/dev/nvidia-uvm
**-** /dev/nvidia-uvm-tools:/dev/nvidia-uvm-tools
Then I also have :
deploy:
resources:
reservations:
devices:
**-** driver: nvidia
count: all
capabilities:
**-** gpu
Nvidia is viewable from the container:
docker info | grep -i nvidia
cdi: nvidia.com/gpu=0
cdi: nvidia.com/gpu=GPU-9995236e-f6d5-4ecf-d087-9fc3e6dbd14c
cdi: nvidia.com/gpu=all
Runtimes: runc io.containerd.runc.v2 nvidia
I verified in my docker daemon.json file that is is forcing legacy mode and not CDI mode…
When I recreate my docker container, and run:
docker exec -it plex ls /usr/lib/x86_64-linux-gnu | grep -i nvidia
I am still missing the HVENC and NVDEC libraries… but I see the display libraries.
I’m not sure, but I think it is still using CDI even though I forced the docker daemon.json and the Nvidia toolkit into legacy mode (I edited the /etc/nvidia-container-runtime/config.toml file and commented out all the CDI block and then changed mode to ‘legacy’)
Does anyone have a working plex container for docker and can help me troubleshoot?
I do not see a GPU widget on the dashboard.
I have enabled HW encoding.
I have manual set the device to the Quadro p600
Under Plex settings, I have set the transcode settings to:
quality to prefer higher quality.
background transcoding: slower
checked the boxes for Use hardware acceleration when available and Use hardware-accelerated video encoding
max gpu transcodes is set to 4