Project template for dockerized Jupyter Lab with AWS

   _         _                 
  | |_ _   _| |_ ___  ___ ___  
  | __| | | | __/ _ \/ __/ _ \ 
  | |_| |_| | ||  __/ (_| (_) |
 (_)__|\__,_|\__\___|\___\___/ 
 
 -- data & knowledge experts --                              

cookiecutter template for data analysis with JupyterLab including a tutorial to get a jump start.

This template is based on the extended docker image tuteco/jupyter_datascience_pyspark. You can see the packages included in the github repository

Usage

Prerequisites:

• Python 3.8+ for your OS
• cookiecutter
• invoke

generate a new project, you simply call

cookiecutter https://github.com/tuteco/cookiecutter-jupyter-aws-docker.git -o <target_directory>

if -o <target directory> is omitted, the project will be generated in the current directory

Parameters you need to provide:

• project_name: short description what you project is about
• project_dir: sluggified version of your project name, or define a directory on your own
• aws_account_number: the account number you are connecting to
• aws_default_region: default region, like eu-central-1
• aws_container_credentials_relative_uri:
  • /creds: use your local credentials from the host OS
  • /role: define a role in AWS
• aws_role: the role name, if /creds chosen before leave it empty

More background information on working with local containers can be found on the AWS Compute Blog

extending the image

For inital working with this setup you don't need to extend the image.

If your requirements grow for additional python libraries or jupyter extension you need for your notebooks, you want to build your own local image. To do this, you need to modify the following files in your project:

• requirements.txt: any python library you want to install
• Dockerfile: any extensions that cannot be installed via pip or other binaries required
• docker-compose.yaml: change the name of the image of the jupyter container

To build the local image, simply run

invoke build-local

The image crated has the default name of local/{project_path}, where the {project_path is the one you specified during the cookiecutter setup. This is the new image name you need to replace for the service container in the in docker-compose.yml.