Search the Community

I've been running Emby with GPU transcoding for multiple years now but if i try to update past version 4.7.6.0 (Linuxserver docker) the container won't even finish the configuration but i can't come here and ask for help with someone elses container so i've been trying to resolve my issues with the official Emby/Embyserver container and i'm hoping for some help The container start fine but nvidia-smi is not found in the shell and the logs are showing: ffdetect_nvencdec Process exited with code 1 nCannot load libcuda.so.1\nError loading CUDA functions\n\n","Result":{ Configuration": "--cc=x86_64-emby-linux-gnu-gcc ...... }, "Error": { "Number": -1, "Message": "Operation not permitted" }, "Log": [ { "Level": 16, "Category": 0, "Message": "Cannot load libcuda.so.1" }, { "Level": 16, "Category": 0, "Message": "Error loading CUDA functions" And clicking advanced under hardware acceleration gives no options Host is Arch Linux 6.1.3-arch1-1, I've been trying to see if this works every few months so it has been tried with older kernels through those months. Emby premiere is enabled Gpu is a quadro p620p NVIDIA-SMI 525.78.01 Driver Version: 525.78.01 CUDA Version: 12.0 docker 1:20.10.22-1 (latest) libnvidia-container 1.5.1-1 libnvidia-container-tools 1.5.1-1 nvidia-container-runtime 3.5.0-2 nvidia-container-toolkit 1.5.1-1 I've previously used the official Emby container without issues but i have never before tried it with hardware encoding For the purpose of finding the issue i've been testing on a fresh install of the official Emby/Embyserver container from hub.docker with premium enabled I've tried other media servers to see if hardware encoding works there and it does I've tried to go backwards with the official Emby docker container but can't seem to find an older version that work with hardware acceleration I've tried to roll back my drivers to an older version with no luck nvidia-smi is not found when run from the containers shell I've seen an issue or two here that resemble mine but they either don't come to a conclusion, Have a different cuda errror or and have not attempted running other containers that requre GPU acceleration, Which works flawlessly for me docker-compose: embytest: image: emby/embyserver:latest hostname: embytest.docker container_name: embytest restart: always networks: - frontend # ports: # - 9096:9096 # - 9920:9920 mem_limit: 4g runtime: nvidia devices: - /dev/dri:/dev/dri deploy: resources: reservations: devices: - capabilities: [gpu] environment: - UID=${PUID} - GID=${PGID} - GIDLIST=985,989 - NVIDIA_VISIBLE_DEVICES=all volumes: - ${DOCKER_HOME}/embytest:/config - media:/mnt/media dns: - 1.1.1.1 - 1.0.0.1 How i configured my host to enable the nvidia-container-runtime Blacklisted nouveau added the kernel parameter (systemd.unified_cgroup_hierarchy=false) edited /etc/nvidia-container-runtime/config.toml (no-cgroups = false) rebuild the boot image added: /etc/docker/daemon.json with: rebooted { "runtimes": { "nvidia": { "path": "/usr/bin/nvidia-container-runtime", "runtimeArgs": [] } } } installed the latest good nvidia driver that works with the nvidia patch patched nvidia rebooted gpu acceleration with the linuxserver docker would not work without this added to docker-compose: deploy: resources: reservations: devices: - capabilities: [gpu] hardware_detection-63808684426.txt embyserver(2).txt

Would anyone be able to help me? I have a docker install for Emby currently working but I would like to use Docker Compose to further simplify the process. I noticed that the Nvidia runtime is not supported within the docker-compose.yml file but does anyone know a way to maybe incorporate the runtime with a Dockerfile? Any help would be appreciated.