Ubuntu Server GPU or non GPU server setup guide using lscripts-docker bash shell script framework
- If there is an error of for
/cdrom
onsudo apt update
, then comment theCDROM
line in the file:/etc/apt/sources.list
cat /etc/apt/sources.list sudo vi /etc/apt/sources.list
- Install ubuntu-desktop
sudo apt -y update sudo apt -y install ubuntu-desktop
- Check the latest nvidia driver version, currently it was 535
sudo apt -y install nvidia-driver-535
- Install minimum server dependencies and lscripts
bash <(curl -s https://raw.githubusercontent.com/skillplot/lscripts-docker/main/lscripts/banners/skplt.serversetup.sh)
- source the bashrc. The CLI prompt will change and will show the skillplot banner, then execute this command
source ~/.bashrc
- Install dependencies by executing following command. NOTE: No need to reboot if it ask for; type
n
. Configure, verify, add repo keys, add repo - whenever it asks for different items.- SKIP: nvidia driver installation
- Install docker-ce
- Install docker-compose
- Install python
- Install python-virtualenvwrapper
- Install nvidia-container-toolkit
- SKIP: cuda stack installation
lsd-stack.nvidia_cuda_python_docker
- Install conda
lsd-install.python-miniconda
- Change the docker subnet if needed; refer following link
- Cleanup
- Remove the configuration lines from the
vi ~/.bashrc
fileexport LSCRIPTS_DOCKER="/tmp/codehub/external/lscripts-docker" [ -f ${LSCRIPTS_DOCKER}/lscripts/lscripts.env.sh ] && source ${LSCRIPTS_DOCKER}/lscripts/lscripts.env.sh
- Remove
/tmp/codehub
directory created by the script containing `external/lscripts-docker from the system.rm -rf /tmp/codehub
- Remove the configuration lines from the