3.2. Running JupyterLab

The following instructions describe how to run the JupyterLab IDE using Docker or Singularity.

Starting the container

Important

Repeat the docker run command if you receive an error like: ‘docker: Error response from daemon’.

Linux or Mac with Docker

  1. Open a new terminal on your host machine.

  2. Check that you are using the bash shell:

echo $SHELL

  1. If the command above does not print /bin/bash then start a bash shell:

bash

  1. Run the following:

docker run \
    --rm \
    --name development-environment \
    -p 10000:8888 \
    -e JUPYTER_ENABLE_LAB=yes \
    -v ~/development-environment/work:/home/jovyan/work \
    -v ~/development-environment/usr/local:/home/jovyan/.local \
    -v ~/development-environment/usr/cache:/home/jovyan/.cache \
    -v ~/development-environment/usr/etc/jupyter:/etc/jupyter \
    -v ~/development-environment/usr/bin/:/usr/local/bin \
    researchdevresources/development-environment:1.0-tensorflow-notebook

Important

Ensure that there are no trailing spaces following the end of line backslash deliminators.

Linux with Singularity

Repeat steps 1-3 above and then run the following:

SINGULARITY_JUPYTER_ENABLE_LAB=yes singularity run \
    -B ~/development-environment/work:/home/jovyan/work \
    -B ~/development-environment/usr/local:/home/jovyan/.local \
    -B ~/development-environment/usr/cache:/home/jovyan/.cache \
    -B ~/development-environment/usr/etc/jupyter:/etc/jupyter \
    -B ~/development-environment/usr/bin/:/usr/local/bin \
    docker://researchdevresources/development-environment:1.0-tensorflow-notebook

Important

Ensure that there are no trailing spaces following the end of line backslash deliminators.

Windows with Docker

Run the following command in a PowerShell (don’t use a standard terminal, as it does not support commands that span multiple lines):

docker run `
    --rm `
    --name development-environment `
    -p 10000:8888 `
    -e JUPYTER_ENABLE_LAB=yes `
    -v c/Users/${env:UserName}/Documents/development-environment/work:/home/jovyan/work `
    -v c/Users/${env:UserName}/Documents/development-environment/usr/local:/home/jovyan/.local `
    -v c/Users/${env:UserName}/Documents/development-environment/usr/cache:/home/jovyan/.cache `
    -v c/Users/${env:UserName}/Documents/development-environment/usr/etc/jupyter:/etc/jupyter `
    -v c/Users/${env:UserName}/Documents/development-environment/usr/bin/:/usr/local/bin/ `
    researchdevresources/development-environment:1.0-tensorflow-notebook

Important

Ensure that there are no trailing spaces following the end-of-line tilda deliminators.

Accessing JupyterLab

The above commands will start a JupyterLab server on port 8888 within the docker container.

A JupyterLab interactive session can be started in the browser of your host machine by copying and pasting the url with the access token (highlighted in yellow in the figure below) into a web browser (e.g. chrome).

../_images/docker_jupyter_server_url.png

Docker JupyterLab server url.

Important

If you are running a Docker container following the instructions below. If you are running a Singularity container then skip this information.

Note that the above url will give a This site can’t be reached 127.0.0.1 refused to connect error. This is because the docker run command above maps port number 8888 within the container to port number 10000 on the host windows machine. Replace 8888 with 10000 in the url and the JupyterLab interactive session will load as expected in your web browser.

Singularity does not require mapping of ports.

Working with JupyterLab

See the IDE page for more information on how to use JupyterLab.