2021-06-21-lscript-docker.md

layout	title	date	categories
post	Lscript Docker	2021-06-21 05:00:25 +0530	lscript-docker

Overview

1. Install nvidia-driver
2. Install docker (community edition), docker-compose (optional) and nvidia-docker-toolkit on the host machine.
3. Build Nvidia CUDA Docker Images
4. CI/CD

Installation

This is for advance usage providing granular control on specific software component that is to be installed.

1. Install Nvidia driver

   lsd-install.nvidia-driver

2. Install docker, docker-compose

   lsd-install.docker-ce
   lsd-install.docker-compose

3. Install Nvidia container toolkit

   lsd-install.nvidia-container-toolkit

Build Nvidia CUDA Docker Images

Only for Nvidia GPU

If looking to use GPU docker containers and have Nvidia Graphics card, multiple CUDA stack can be installed and used in parallel using Nvidia GPU docker containers.

These provides complete development stack for deep learning docker environments. Any AI frameworks can be added on top of it and currently provides pre-configuration for Pytorch, Keras and Tensorflow AI frameworks.

• Pre-requites
  • Nvidia Driver
  • Docker
  • Nvidia runtime (or nvidia container toolkit which comes with the runtime)
• Build CUDA 10.0 Nvidia image

  ## create the external directory, inside this clone the official nividia container file repo
  mkdir -p external
  cd external
  git clone https://gitlab.com/nvidia/container-images/cuda.git
  ##
  printf "y" | bash docker-buildimg-cuda.sh 10.0

• Python dependencies for specific CUDA versions

  pip install -U -r lscripts/config/ubuntu18.04/python.requirements-ai-cuda-9.0.txt
  pip install -U -r lscripts/config/ubuntu18.04/python.requirements-ai-cuda-10.0.txt
  pip install -U -r lscripts/config/ubuntu18.04/python.requirements-ai-cuda-10.2.txt
  #
  pip install -U -r lscripts/config/ubuntu18.04/python.requirements-ai-cuda-11.0.txt

• Docker container tips
  • Re-start host docker service if Internet access is not there inside container;

    sudo service docker stop
    sudo service docker start
    #
    bash start.bzohub-170820_051654.sh
    #
    docker exec -u $(id -u):$(id -g) -it c1c80473e63c /bin/bash && xhost -local:root 1>/dev/null 2>&1
    docker exec -u $(id -u):$(id -g) -it bzohub-170820_051654 /bin/bash && xhost -local:root 1>/dev/null 2>&1

lsd-docker Commands

• Quick Commands:

  lsd-docker.osvers
  #
  lsd-docker.cfg
  lsd-docker.container.delete-all
  lsd-docker.container.delete-byimage
  lsd-docker.container.exec-byname
  lsd-docker.container.list
  lsd-docker.container.list-ids
  lsd-docker.container.list-ids-all
  lsd-docker.container.status
  lsd-docker.container.stop-all
  lsd-docker.container.test