docker install

Install git, docker and docker-compose.
Note: don't forget docker's post-installation steps! (eg for linux).

Fetch the source:

$ git clone https://gitlab.inria.fr/scripta/escriptorium.git

Copy the environement variables file

$ cd escriptorium && cp variables.env_example variables.env  

Change some of those variables if you wish to,
notably the super-user username and password are created from DJANGO_SU_NAME and DJANGO_SU_PASSWORD, defaulting to admin:admin.

Build and run the docker containers

$ docker-compose up -d --build  

You should be able to access the website at http://localhost:8080/

To update:

$ git pull  
$ docker-compose up -d --build  

Use your GPU:

1. Install cuda & cudnn
2. Create a file named local.yml with this content

version: "2.4"

services:
   celery-gpu:
      shm_size: '3gb'
      runtime: nvidia
      environment:
        - KRAKEN_TRAINING_DEVICE=cuda:0
        - NVIDIA_VISIBLE_DEVICES=all
        - NVIDIA_DRIVER_CAPABILITIES=all

3. When you build use both yml files

$ docker-compose -f docker-compose.yml -f local.yml up -d --build