Ubuntu 20.04에서 deeptream dGPU 환경 docker 설치 Cheat Sheet

dGPU 환경 기준, Jetson 환경 기준은 별도 Guide 참고 필요.

Deepstream 6.2 기준

https://resources.nvidia.com/en-us-deepstream-get-started-with-c-cpp

DeepStream SDK Development Guide

 

Docker-ce 설치

 $ sudo apt-get update
 $ sudo apt-get install apt-transport-https ca-certificates
 $ curl gnupg-agent software-properties-common
 $ curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
 $ sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu bionic stable"
 $ sudo apt-get update
 $ sudo apt-get install docker-ce

 

Nvidia-docker 설치

  $ curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
   $ distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
   $ curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
   $ sudo apt-get update
   $ sudo apt-get install nvidia-docker2
   $ sudo systemctl restart docker.service
   $ sudo usermod -a -G docker <<USER NAME>>
   $ reboot

  

사용자 설치 스크립트 실행

 $ cd /opt/nvidia/deepstream/deepstream-6.2/user_additional_install.sh

 

Conatinaer 내에 미설치 패키지 설치

  1. cuda-toolkit

 $ apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/3bf863cc.pub
 $ add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /"
 $ apt-get update
 $ apt-get install cuda-toolkit-11-8
 $ apt install cuda-toolkit-11-8

 

Docker container 내부에서 X11이 가능하게 container 실행 방법

1. One-shot

$ sudo nvidia-docker run -it --rm --net=host -v ${HOME}:${HOME}/home -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --volume="$HOME/.Xauthority:/root/.Xauthority:rw" <<<Image Name or Checksum>>> /bin/bash 

2. Container 생성

$ sudo nvidia-docker run -it --net=host -v ${HOME}:${HOME}/home -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --volume="$HOME/.Xauthority:/root/.Xauthority:rw" --name=deepstream-6.2-devel-temp <<<Image Name or Checksum>>> /bin/bash

 아래는 nvidia sample command

# Pull the required docker.  Refer Docker Containers table to get docker container name.
$ docker pull <required docker container name>
# Step to run the docker
$ export DISPLAY=:0
$ xhost +
$ docker run -it --rm --net=host --gpus all -e DISPLAY=$DISPLAY --device /dev/snd -v /tmp/.X11-unix/:/tmp/.X11-unix <required docker container name>