Ganimede is a Jupyter Notebook server that contains additional Jupyter tools and libraries in order to get a more complete solution without the hassle of installing and configuring them. Ganimede can be used either as a Docker container or inside conda environments. For more details on how to setup Ganimede in conda environment look at the section below.

To run the Ganimede docker container:

make docker-pull
make docker-run

Go to http://localhost:8888 to access to the JupyterLab.

The docker image is built on top of all-spark-notebook image. The following are the package/tools provided by Ganimede apart from the ones already available in all-spark-notebok.

• Almond
• DrawIO
• IPySheet
• JupyText
• Jupyter Contrib extensions
• Jupyter Vim binding
• JupyterLab Debugger
• JupyterLab Git
• JupyterLab Github
• JupyterLab TOC
• JupyterLab Vim
• Livy
• Papermill
• Qgrid
• S3Contents
• SparkMagic
• Voila

• dask
• joblib
• koalas
• luigi
• pyarrow
• pyathena
• tqdm

• altair
• ipysheet
• qgrid

• gensim
• gluon-nlp
• nltk
• spacy
• textdistance

• dgl
• gluon-cv
• keras
• mxnet
• pymc3
• shap
• tensorflow
• xgboost

To pull Ganimede from Docker Hub:

make docker-pull

To build the Ganimede Docker image:

make docker-build

To remove an old image:

# Get list of all existing images
docker images
# To clean up disk with old images
docker image rm <name-image>

To see all containers (even the ones no longer running):

docker ps -a

To kill a running container (for gracefully stop use stop command instead):

docker kill <container-id>

To remove old containers (use --rm when using run command to avoid this annoying step):

docker rm <container-id>

To run the Docker container:

make docker-run WORK_DIR="/path/to/workdir"

To run the Docker container in background:

make docker-background-run WORK_DIR="/path/to/workdir"

For more information take a look here.

To automatically start container at boot time:

make docker-boot-run WORK_DIR="/path/to/workdir"

For more information take a look here.

The folder .jupyter contains the jupyter settings. Add the following mount when running the container, considering your settings are located in /home/username/jupyter-config:

     -v /home/username/jupyter-config/bash_history:/home/jovyan/.bash_history \
     -v /home/username/jupyter-config/jupyter:/home/jovyan/.jupyter

If your system supports systemd you can run Ganimede as a Systemd service

make start-systemd-service WORK_DIR="/path/to/workdir"

sudo systemctl restart docker

To run a shell in the Docker container:

make docker-shell

docker system prune --all --volumes

There are scenarios where the Jupyter tools and the python libraries are in different conda environments. You can use the scripts in bin/ directory to setup specific set of packages depending on the environment.

For example to setup the Ganimede tools in the conda environment jupyter-server and setup python libraries to myenv:

make -f Makefile-conda conda-setup-tools CONDA_ENV=jupyter-server
make -f Makefile-conda conda-setup-libraries CONDA_ENV=myenv