Burning Books - SSH Driven Spark Notebook

Burning Books is a web notebook that conveys it commands to a cluster thorugh a SSH connection. It uses the default SSH autentication to control access.

It is in alpha right now, use at your own risk.

Dependences

This image depends on a mongodb instance running somewhere and an SSH host with spark installed to work.

Running

There is a couple of ways that you can run this application. By default it poinst to the port 9085.

With docker run

You can run using docker run and using the enviroment variables MONGO and SPARK_HOST.

docker run -d -p 9085:9085 -e MONGO=[mongodb connection url] -e SPARK_HOST=[spark host] --name burning-books tcoutinho/burning-books

With a cocker-compose file

Docker compose let's you download and run a mongodb instance along this one in a neat service package.

docker-compose.yml

version: '3'
services:
    web:
        image: tcoutinho/burning-books
        ports:
            - 9085:9085
        links:
            - mongo
        restart: always
        environment:
            MONGO: mongodb://burning:book@mongo:27017
            SPARK_HOST: [your host here]

    mongo:
        image: mongo
        restart: always
        ports:
            - 27017:27017
        enviroment:
            MONGO_INITDB_ROOT_USERNAME: burning
            MONGO_INITDB_ROOT_PASSWORD: book

From github repository

If you are felling adventurous and want a local running application:

1. Clone the repository from github
1. Set your enviroment variables on the `production.env` file
1. Run `npm run build` to create the scafolding for the application (it only needs
to be run once)
1. Run `npm start` to start

When the need to update rises, update the application with a git pull and start from step 2.

If you want to debug or develop, you can skip all the steps after cloning the project, install dependencies with npm i in the root directory and client directory and run the application in development mode with npm run dev

Connecting with identity files

If you need a identity file to login in your spark host (AWS users, for example), set the enviroment variable LOGIN_TYPE to SSH. Changing it to SSH will make the login screen ask for your identity file during the login.

Enviroment Variables

Variable	Default	Meaning
LOG_LEVEL	INFO	Level of the log messages
SPARK_HOST	localhost	SSH host with spark-shell installed
MONGO	mongodb://localhost:27017	Mongodb connection string
USER_BLACKLIST		Users to be denied access
SPARK_QUEUE		Default queue of all new spark-shell sessions
SPARK_LIBRARIES		Libraries in SPARK_HOST to be used
LOGIN_TYPE	PASSWORD	Type of login, between PASSWORD and SSH

Editor

We are using Ace as the code editor. It is the Cloud 9 editor and comes pre packaged with Scala, so it was an obvious choice.

For a list of the default keyboard shortcuts visit here.