hannahwoodward / docker-jupyter-climate-earth

Docker image that loads JupyterHub and some additional packages via conda

Jupyter & Earth Climate Docker Image

Docker image to load Jupyter and Conda, with a few additional packages installed useful for climatological data analysis.

Additional Earth-specific packages have been installed for downloading from the Climate Data Store (cdsapi), esgf-pyclient for opendap authentication (e.g. for accessing CORDEX data), and regionmask for CMIP6/IPCC and other region masks. For general geospatial/climate data analysis, see hannahwoodward/docker-jupyter-climate

Useful links

• Jupyter Docker stacks github
• Jupyter Docker stacks docs
• Docker build help
• Docker run help

Installation & Use

• Install Docker desktop
• Ensure Docker desktop is running
• Navigate to your project directory (i.e. with Jupyter notebook files)
• Download published image:

docker pull woodwardsh/jupyter-climate-earth:latest

• Run published image, mounting local working directory to container directory /home/jovyan:

docker run -it --rm -p 8888:8888 -v ${PWD}:/home/jovyan -w /home/jovyan woodwardsh/jupyter-climate-earth:latest

# Options:
# -it           interactive && TTY (starts shell inside container)
# --rm          delete container on exit
# --volume|-v   mount local directory inside container
# -w PATH       sets working directory inside container

Exporting the Conda environment

• Start the container
• Open a new terminal inside JupyterLab
• Run sh ~/generate-config.sh to generate config climate-earth.yml into your working directory
• Download climate-earth.yml from the JupyterLab UI, or copy straight from the directory if mounted

Building & running image from scratch

• Clone repo & navigate inside:

git clone [email protected]:hannahwoodward/docker-jupyter-climate-earth.git && cd docker-jupyter-climate-earth

• Build image from Dockerfile (takes ~15 minutes):

docker build -t jupyter-climate-earth .

# -t = name/tag the image, format `name:tag`

• Navigate to your project directory (i.e. with Jupyter notebook files)
• Run locally built image, mounting local working directory to container directory /home/jovyan:

docker run -it --rm -p 8888:8888 -v ${PWD}:/home/jovyan -w /home/jovyan jupyter-climate-earth

# Options:
# -it           interactive && TTY (starts shell inside container)
# --rm          delete container on exit
# --volume|-v   mount local directory inside container
# -w PATH       sets working directory inside container

Publishing image

docker login && docker tag jupyter-climate-earth woodwardsh/jupyter-climate-earth && docker push woodwardsh/jupyter-climate-earth

Troubleshooting

• Exit code 137 - need to increase Docker memory e.g. to 4GB
• No space left on device - docker system prune