This is the first sample container for which we demonstrate how to use some HPC tooling on Kubernetes or locally, using the https://cgs.jhuapl.edu codebase "MAGE" ( an ecosystem consisting of Fortran/Python MPI/OPENMP). Here, we specifically use Paraview to plot Magneto-Hydro-Dynamics Data produced by MAGE, and thus we include the Python Packages required specifically for this HPC codebase (kaipy etc).
WIP: The Globus Endpoints for the data are public, but currently those datasets are rather large.
For a basic example of the kernel using Jupyter Notebook, see ParaviewExample.ipynb, for a sample Tutorial see MagIonPlot.ipynb , for opening a connection via Globus, see globus_client_start.ipynb. To have more information on what is available specifically through this Jupyter-kernel, type magic command %help in the interpreter.
docker run -it -p 8888:8888 ghcr.io/austriandatalab/aocc_openscience_mage:main
and you can open a MAGE tutorial in Jupyter and have the Paraview-Kernel to visualize HDF5 data, accessible in your browser.
You can also see our other repos for the full JupyterHub on Docker and Minikube (https://github.com/AustrianDataLAB/AOCC-JupyterHub-Trial)
Included you find a Dockerfile with a simple test environment based on JupyterLab.
The installation is based on Python 3.9 and uses miniconda to install Paraview 5.11.1 (on conda-forge).
If you wish to add your own modifications, clone this repo, modify it, then:
To build the container run:
$ docker build -t my_openscience_paraview:0.0.1 .and to run it:
$ docker run -it -p 8888:8888 my_openscience_paraview:0.0.1You will see the URL showing up to connect to the JupyterLab.
see our main project under: (https://dev.azure.com/AOCC/OpenScienceLabs)