Creating your own JuypterHPC container

Users with more complex requirements can build their own Apptainer containers and load them in JupyterHPC to be able to access them on the web.

The following basic container recipe can be used as a starting point. It is important to keep the version of the jupyterhub package close to the one that is currently used on the hub (4.1.5).

Bootstrap: docker
From: condaforge/miniforge3

%post
    export DEBIAN_FRONTEND=noninteractive
    apt update
    apt upgrade -y

    conda install --quiet --yes \
        'ipyparallel' \
        'notebook=7.2.1' \
        'jupyterhub=4.1.5' \
        'jupyterlab=4.2.3'

    # Here you can add your own python packages
    conda install --quiet --yes \
        pandas \
        scipy \
        seaborn

A jupyterhub-singleuser binary that is compatible with the hub must be in the $PATH of the container for a successful startup. Otherwise you can extend the container as much as you want.

A more complex example that includes RStudio integration as well as PyTorch is shown below:

Bootstrap: docker
From: condaforge/miniforge3

%post
    export DEBIAN_FRONTEND=noninteractive 
    apt update 
    apt upgrade -y

    apt install -y --no-install-recommends software-properties-common dirmngr
    wget -qO- https://cloud.r-project.org/bin/linux/ubuntu/marutter_pubkey.asc | tee -a /etc/apt/trusted.gpg.d/cran_ubuntu_key.asc
    add-apt-repository "deb https://cloud.r-project.org/bin/linux/ubuntu $(lsb_release -cs)-cran40/"
    add-apt-repository ppa:c2d4u.team/c2d4u4.0+
    apt update

    apt install -y \
        r-base \
        r-cran-caret \
        r-cran-crayon \
        r-cran-devtools \
        r-cran-forecast \
        r-cran-hexbin \
        r-cran-htmltools \
        r-cran-htmlwidgets \
        r-cran-plyr \
        r-cran-randomforest \
        r-cran-rcurl \
        r-cran-reshape2 \
        r-cran-rmarkdown \
        r-cran-rodbc \
        r-cran-rsqlite \
        r-cran-shiny \
        r-cran-tidyverse 

    apt install -y libclang-dev lsb-release psmisc sudo
    ubuntu_release=$(lsb_release --codename --short) 
    wget https://download2.rstudio.org/server/${ubuntu_release}/amd64/rstudio-server-2023.12.1-402-amd64.deb
    dpkg --install rstudio-server-2023.12.1-402-amd64.deb
    rm rstudio-server-2023.12.1-402-amd64.deb

    echo 'ftp_proxy=http://www-cache.gwdg.de:3128' >> /usr/lib/R/etc/Renviron.site
    echo 'https_proxy=http://www-cache.gwdg.de:3128' >> /usr/lib/R/etc/Renviron.site
    echo 'http_proxy=http://www-cache.gwdg.de:3128' >> /usr/lib/R/etc/Renviron.site

    echo '' >> /usr/lib/R/etc/Renviron.site
    conda install --quiet --yes \
        'ipyparallel=8.8.0' \
        'jupyter-rsession-proxy=2.2.1' \
        'notebook=7.2.1' \
        'jupyterhub=4.1.5' \
        'jupyterlab=4.2.3'
   
    conda install --quiet --yes \
        dgl \
        igraph \
        keras \
        pandas \
        pydot \
        scikit-learn \
        scipy \
        seaborn

For HPC Desktops, even the smallest example would consist of over 100 lines, thus they will not be repeated here. Please take your inspiration from the original recipes under /sw/viz/jupyterhub-nhr/juypter-containers/