README

Dockerfiles used for our workstations and user specific dockerfiles.

Root Image

Define :

* PORT (port to use for ssh)
* HOME MOUNT (mounts and backs up home folder)
* SSD MOUNT (mounts SSD for binaries)
* D_USERNAME 

Run

nvidia-docker run -it -d --name $D_USERNAME \
-p $PORT:22 \
-e USERNAME=$D_USERNAME \
-v $HOME_MOUNT$:/home/$D_USERNAME \
-v $SSD_MOUNT:/mnt/ssd \
iitmcvg/workstation bash

NGC- python3 Tensorflow based.

Required nvidia-docker for GPU runs.

Tested with nvidia-drivers (384,387,390,394)

Packages: * sudo
* git
* wget
* zsh
* vim
* zsh
* htop
* imagemagick
* fdupes
* libcurl4-gnutls-dev
* wget
* openssh-server
* curl

Entrypoint script

Sets username, passwords.

Recommended to change from default to SSH-Keys.

Also exports HOME, PATH, LD_LIBRARY_PATH.

Running User-Specific Images

Replace iitmcvg/workstation by iitmcvg/workstation#master:dir.

Can use other branches if available.

Run (eg):

nvidia-docker run -it --rm --name $USERNAME \
-p 18888:8888 -p 1002:22 -p 16006:6006 -p 1010-1100:1010-1100 \
-e USERNAME=$USERNAME \
-v /media/home/$USERNAME$:/home/$USERNAME \
-v /media/ssd/$USERNAME$:/mnt/ssd \
iitmcvg/workstation/#master:torch bash