Solving environment.. working.. failed on Ubuntu 20.04.1 LTS

[sushrestha]

I am unable to build the Jupyterlab image on Ubuntu 20.04.1 LTS. Tried adding channels on Conda config but no success. Do you think I have to work with the latest version of packages? Is there any guidelines or document for the new version of Jupyterhub docker image?

[tony-sappe]

@sushrestha I'm curious to learn more about the issue. Am I correct inferring you're experiencing issues with the conda install steps?

I updated the scipy-notebook image to be a little newer, but still an older version due to the packages I'm working with. Even the image is newer, the packages are still very old. Through a long day of waiting I realized the install steps needed to be separate from update commands. There was simply too many dependency trees to traverse for the package manager to quickly resolve all versions.

Here's what I have in my Dockerfile for JupyterLab. Let me know if you already solved the issue another way or if this helped.

# https://jupyter-docker-stacks.readthedocs.io/en/latest/using/selecting.html#jupyter-scipy-notebook
FROM jupyter/scipy-notebook:f9c990eb6295

USER root

# APT packages
RUN apt-get update \
    && apt-get upgrade -y \
    && apt-get install -y --no-install-recommends \
    fonts-dejavu \
    tzdata \
    gfortran \
    gcc \
    scilab \
    pari-gp \
    libpari-dev \
    libgmp-dev \
    && apt-get autoremove \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

USER $NB_UID

RUN conda config --add channels conda-forge \
  && conda config --set channel_priority strict

RUN conda update -n base conda
RUN conda update --all
RUN conda install --quiet --yes \
    <YOUR EXTRA PACKAGES HERE>
RUN conda clean --all --yes \
    && fix-permissions $CONDA_DIR

ENV CPATH=$CONDA_DIR/include

RUN pip install \
    pari_jupyter \
    scilab-kernel \
    && fix-permissions $CONDA_DIR

# Add conda env hook
COPY ./conda-activate.sh /usr/local/bin/before-notebook.d/