patelravip/jupyterlab repository overview

⁠JupyterLab with Python 2 and 3 Kernels for Docker

Docker image for a JupyterLab instance that includes Python 2 and 3 kernels as well as associated packages required for notebook widgets.

⁠Relevant links

• GitHub @ rpatel/docker-jupyterlab⁠
• Docker Registry @ patelravip/jupyterlab⁠
• JupyterLab @ JupyterLab Documentation⁠

⁠Quickstart

⁠Fetch the image

• Pull the latest automated build using the docker pull command: docker pull patelravip/jupyterlab OR clone the repository and build a local image: docker build -t my-jupyter-image

⁠Collect your requirements

• Make text files with python requirements that will be installed by pip on start:

  e.g., my_python2_requirements.txt and/or my_python3_requirements.txt:

  requests
  bs4
  
  numpy
  scipy
  pandas
  matplotlib
  
  stevedore>=1.3.0,<1.4.0
  

  Copy

  following pip requirments file format⁠.

  Note: requirements can be installed from inside JupyterLab as well; this is just for set-and-forget convenience when moving JupyterLab setups around.

⁠Generate basic notebook configuration file

• Run jlab_genconfig command to generate initial configuration file if you do not already have one.

  $ touch /path/to/my/new/config/file/jupyter_notebook_config.py
  
  $ sudo chown :100000 /path/to/my/new/config/file/jupyter_notebook_config.py
  
  $ chmod g+rw /path/to/my/new/config/file/jupyter_notebook_config.py
  
  $ docker run --rm --log-driver=none -ti \
      -v "/path/to/my/new/config/file/jupyter_notebook_config.py:/home/jupyter/.jupyter/jupyter_notebook_config.py" \
      patelravip/jupyterlab jlab_genconfig
  

  Copy

  If you do not have sudo permissions, you can expose the file to all other users instead of sudo chown ... above:

  $ chmod o+rw /path/to/my/new/config/file/jupyter_notebook_config.py
  

  Copy

⁠Make a directory to store notebooks

• Make a directory for Notebooks that is writable by a UID/GID 100000 (jupyterlab user in container):

  $ mkdir /path/to/my/notebooks/directory
  
  $ sudo chown :100000 /path/to/my/notebooks/directory
  
  $ chmod g+rw /path/to/my/notebooks/directory
  

  Copy

  If you do not have sudo permissions, you can expose the directory to all other users instead of sudo chown above:

  $ chmod o+rw /path/to/my/notebooks/directory
  

  Copy

⁠Finally,

• Start JupyterLab on Port 8888:

  $ docker run --log-driver=none \
      --name my_first_lab_container \
      -v "/path/to/my_python2_requirements.txt:/requirements.py2.txt" \
      -v "/path/to/my_python3_requirements.txt:/requirements.py3.txt" \
      -v "/path/to/my/new/config/file/jupyter_notebook_config.py:/home/jupyter/.jupyter/jupyter_notebook_config.py" \
      -v "/path/to/my/notebooks/directory:/home/jupyter/Notebooks \
      -p "8888:8888" \
      patelravip/jupyterlab
  

  Copy

  Modify the volumes, -v, according to your requirement files, config file, and Notebooks directory.

  You should be able to access JupyterLab from your browser at the following address: http://localhost:8888/⁠. The token for login will be provided in the log messages in the terminal after executing the command above.

⁠Further configuration

⁠Use a password instead of a token for login

If you used the link with token from the startup log, you bypassed the screen to set a password to avoid using token authentication in the future. You can still manually set a password.

• Use the jupyter notebook password command to set a password that will be saved as a hash in your configuration file:

  $ docker run --rm --log-driver=none \
      -v "/path/to/my/new/config/file/jupyter_notebook_config.py:/home/jupyter/.jupyter/jupyter_notebook_configuration.py" \
      patelravip/jupyterlab jupyter notebook password
  

  Copy

  After running the command to start your JupyterLab instance, you will now be prompted for a password instead of having to use the link or provide the randomized token from the startup log.

⁠Adding OS level packages

Often you will need to use a system executable in a python script, e.g. with subprocess, install a package that is required by a python module, or install a python package not available with pip but is in the Debian repositories (base image is currently based on Debian stretch).

• Use docker exec to execute commands as root in your running container:

  • Update the package list:

    $ docker exec --user=root my_first_lab_container apt-get update
    

    Copy

  • Install the desired packages (e.g., tabix and pymol)

    $ docker exec --user=root my_first_lab_container apt-get -y install tabix pymol
    

    Copy

  Note: replace my_first_lab_container with the container name you set when you first started your JupyterLab instance.

⁠Additional reading

• Read the JupyterLab Documentation⁠
• Read about securing your notebook communication⁠ to avoid things like sending your hashed password unencrypted by your browser.
• Check out additional notebook configuration options⁠ to further customize your jupyter_notebook_config.py file (edit manually)