Jupyter

From CC Doc
Jump to: navigation, search
This page contains changes which are not marked for translation.

Other languages:
English • ‎français

Introduction

"Project Jupyter is a non-profit, open-source project, born out of the IPython Project in 2014 as it evolved to support interactive data science and scientific computing across all programming languages."[1]

"The Jupyter Notebook is an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text."[2]

You can run Jupyter Notebook on a compute node or on a login node (not recommended). Note that login nodes impose various user- and process-based limits, so applications running there may be killed if they consume too much CPU time or memory. To use a compute node you will have to submit a job requesting the number of CPUs (and optionally GPUs), the amount of memory, and the run time. Here, we give instructions to submit a Jupyter Notebook job.

Some regional partners provide a web portal named JupyterHub so that users do not have to create their own Jupyter Notebook setup. To learn more, visit the JupyterHub wiki page.

Installing Jupyter Notebook

These instructions install Jupyter Notebook with the pip command in a Python virtual environment in your home directory. The following instructions are for Python 3.6, but you can also install the application for a different version by loading a different Python module.

  1. Load the Python module.
    [name@server ~]$ module load python/3.6
    
  2. Create a new Python virtual environment.
    [name@server ~]$ virtualenv $HOME/jupyter_py3
    
  3. Activate your newly created Python virtual environment.
    [name@server ~]$ source $HOME/jupyter_py3/bin/activate
    
  4. Install Jupyter Notebook in your new virtual environment.
    (jupyter_py3)_[name@server ~]$ pip install jupyter
    
  5. In the virtual environment, create a wrapper script that launches Jupyter Notebook.
    (jupyter_py3)_[name@server ~]$ echo -e '#!/bin/bash\nunset XDG_RUNTIME_DIR\njupyter notebook --ip $(hostname -f) --no-browser' > $VIRTUAL_ENV/bin/notebook.sh
    
  6. Finally, make the script executable.
    (jupyter_py3)_[name@server ~]$ chmod u+x $VIRTUAL_ENV/bin/notebook.sh
    

Installing extensions

Extensions allow you to add functionalities and modify the application’s user interface.

Jupyter Lmod

Jupyter Lmod is an extension that allows you to interact with environment modules before launching kernels. The extension uses the Lmod's Python interface to accomplish module-related tasks like loading, unloading, saving a collection, etc.

(jupyter_py3)_[name@server ~]$ pip install jupyterlmod
(jupyter_py3)_[name@server ~]$ jupyter nbextension install --py jupyterlmod --sys-prefix
(jupyter_py3)_[name@server ~]$ jupyter nbextension enable --py jupyterlmod --sys-prefix
(jupyter_py3)_[name@server ~]$ jupyter serverextension enable --py jupyterlmod --sys-prefix

Proxy web services

nbserverproxy enables users to reach arbitrary web services running within their spawned Jupyter server. This is useful to access web services that are listening only on a port of the localhost like TensorBoard.

(jupyter_py3)_[name@server ~]$ pip install nbserverproxy
(jupyter_py3)_[name@server ~]$ jupyter serverextension enable --py nbserverproxy --sys-prefix

Example

In Jupyter, a user starts a web service via 'Terminal' in the New dropdown list:

[name@server ~]$ tensorboard --port=8008

The service is proxied off of /proxy/ at https://address.of.notebook.server/user/theuser/proxy/8008.

RStudio Launcher

Jupyter Notebook can start an RStudio session that uses Jupyter Notebook's token authentication system. RStudio Launcher adds an RStudio Session option to the Jupyter Notebook New dropdown list.

(jupyter_py3)_[name@server ~]$ pip install nbserverproxy
(jupyter_py3)_[name@server ~]$ pip install git+https://github.com/jupyterhub/nbrsessionproxy
(jupyter_py3)_[name@server ~]$ jupyter serverextension enable --py nbserverproxy --sys-prefix
(jupyter_py3)_[name@server ~]$ jupyter nbextension install --py nbrsessionproxy --sys-prefix
(jupyter_py3)_[name@server ~]$ jupyter nbextension enable --py nbrsessionproxy --sys-prefix
(jupyter_py3)_[name@server ~]$ jupyter serverextension enable --py nbrsessionproxy --sys-prefix


Activating the environment

Once you have installed Jupyter Notebook, you need only reload the Python module associated with your environment when you log into the cluster.

[name@server ~]$ module load python/3.6

Then, activate the virtual environment in which you have installed Jupyter Notebook.

[name@server ~]$ source $HOME/jupyter_py3/bin/activate

RStudio Server (optional)

To use RStudio Launcher, load the RStudio Server module.

(jupyter_py3)_[name@server ~]$ module load rstudio-server

Starting Jupyter Notebook

To start the application, submit an interactive job. Adjust the parameters based on your needs. See Running jobs for more information.

(jupyter_py3)_[name@server ~]$ salloc --time=1:0:0 --ntasks=1 --cpus-per-task=2 --mem-per-cpu=1024M --account=def-yourpi srun $VIRTUAL_ENV/bin/notebook.sh
salloc: Granted job allocation 1422754
salloc: Waiting for resource configuration
salloc: Nodes cdr544 are ready for job
[I 14:07:08.661 NotebookApp] Serving notebooks from local directory: /home/fafor10
[I 14:07:08.662 NotebookApp] 0 active kernels
[I 14:07:08.662 NotebookApp] The Jupyter Notebook is running at:
[I 14:07:08.663 NotebookApp] http://cdr544.int.cedar.computecanada.ca:8888/?token=7ed7059fad64446f837567e32af8d20efa72e72476eb72ca
[I 14:07:08.663 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 14:07:08.669 NotebookApp]

Copy/paste this URL into your browser when you connect for the first time,
    to login with a token:
        http://cdr544.int.cedar.computecanada.ca:8888/?token=7ed7059fad64446f837567e3

Connecting to Jupyter Notebook

To access Jupyter Notebook running on a compute node from your web browser, you will need to create an SSH tunnel between the cluster and your computer since the compute nodes are not directly accessible from the Internet.

From Linux or MacOS X

On a Linux or MacOS X system, we recommend using the sshuttle Python package.

On your computer, open a new terminal window and run the following sshuttle command to create the tunnel.

[name@my_computer ~]$ sshuttle --dns -Nr <username>@<cluster>.computecanada.ca

In the preceding command substitute <username> by your Compute Canada username; and substitute <cluster> by the cluster you connected to launch your Jupyter Notebook.

Then, copy and paste the provided URL into your browser. In the above example, this would be

 http://cdr544.int.cedar.computecanada.ca:8888/?token=7ed7059fad64446f837567e3

From Windows

An SSH tunnel can be created from Windows using MobaXTerm as follows. This will also work from any Unix system (MacOS, Linux, etc).

  1. Open a new Terminal tab in MobaXTerm (Session 1) and connect to a cluster. Then follow the instructions in section Starting Jupyter Notebook. At this point, you should have on your screen an URL with the following form.
    http://cdr544.int.cedar.computecanada.ca:8888/?token=7ed7059fad64446f837567e3
           └────────────────┬───────────────────┘        └──────────┬───────────┘
                      hostname:port                               token
    
  2. Open a second Terminal tab in MobaXTerm (Session 2). In the following command, substitute <hostname:port> by its corresponding value from the URL you obtained in Session 1 (refer to the previous figure); substitute <username> by your Compute Canada username; and substitute <cluster> by the cluster you connected to in Session 1. Run the command.
    [name@my_computer ~]$ ssh -L 8888:<hostname:port> <username>@<cluster>.computecanada.ca
    
  3. Open your browser and go to
     http://localhost:8888/?token=<token>
    

    Replace <token> with its value from Session 1.

Shutting down Jupyter Notebook

You can shut down the Jupyter Notebook server before the walltime limit by pressing Ctrl-C twice in the terminal that launched the interactive job.

If you used MobaXterm to create a tunnel, press Ctrl-D in Session 2 to shut down the tunnel.


Adding kernels

It is possible to add kernels for other programmming languages or Python versions different than the one running the Jupyter Notebook. Refer to Making kernels for Jupyter to learn more.

The installation of a new kernel is done in two steps.

  1. Installation of the packages that will allow the language interpreter to communicate with Jupyter Notebook.
  2. Creation of a file that will indicate to Jupyter Notebook how to initiate a communication channel with the language interpreter. This file is called a kernel spec file.

Each kernel spec file has to be created in its own subfolder inside a folder in your home directory with the following path ~/.local/share/jupyter/kernels. Jupyter Notebook does not create this folder, so the first step in all cases is to create it. You can use the following command.

[name@server ~]$ mkdir -p  ~/.local/share/jupyter/kernels

In the following sections, we provide a few examples of the kernel installation procedure.

Anaconda

Before installing an Anaconda kernel, make sure you have read the documentation and installed Anaconda.

  1. Load the Anaconda module.
    [name@server ~]$ module load miniconda3
    
  2. Optional: Activate a specific conda virtual environment. Substitute <your_env> by the name you want to attribute to your Anaconda virtual environment.
    [name@server ~]$ source activate <your_env>
    
  3. Install the ipykernel library.
    [name@server ~]$ conda install ipykernel
    
  4. Generate the kernel spec file. Substitute <unique_name> by a name that will uniquely identify your kernel.
    [name@server ~]$ python -m ipykernel install --user --name <unique_name> --display-name "My Anaconda 3 Kernel"
    
  5. Optional: Deactivate the virtual environment.
    [name@server ~]$ source deactivate
    

For more information, see the ipykernel documentation.

Julia

  1. Load the Julia module.
    [name@server ~]$ module load julia
    
  2. Activate the Jupyter Notebook virtual environment.
    [name@server ~]$ source $HOME/jupyter_py3/bin/activate
    
  3. Install IJulia.
    (jupyter_py3)_[name@server ~]$ echo 'Pkg.add("IJulia")' | julia
    

For more information, see the IJulia documentation.

Python

  1. Load the Python module.
    [name@server ~]$ module load python/3.5
    
  2. Create a new Python virtual environment.
    [name@server ~]$ virtualenv $HOME/jupyter_py3.5
    
  3. Activate your newly created Python virtual environment.
    [name@server ~]$ source $HOME/jupyter_py3.5/bin/activate
    
  4. Install the ipykernel library.
    [name@server ~]$ pip install ipykernel
    
  5. Generate the kernel spec file. Substitute <unique_name> by a name that will uniquely identify your kernel.
    [name@server ~]$ python -m ipykernel install --user --name <unique_name> --display-name "Python 3.5 Kernel"
    
  6. Deactivate the virtual environment.
    [name@server ~]$ source deactivate
    

For more information, see the ipykernel documentation.


R

  1. Load the R module.
    [name@server ~]$ module load r
    
  2. Activate the Jupyter Notebook virtual environment.
    [name@server ~]$ source $HOME/jupyter_py3/bin/activate
    
  3. Install the R kernel dependencies.
    (jupyter_py3)_[name@server ~]$ R -e "install.packages(c('crayon', 'pbdZMQ', 'devtools'), repos='http://cran.us.r-project.org')"
    
  4. Install the R kernel.
    (jupyter_py3)_[name@server ~]$ R -e "devtools::install_github(paste0('IRkernel/', c('repr', 'IRdisplay', 'IRkernel')))"
    
  5. Install the R kernel spec file.
    (jupyter_py3)_[name@server ~]$ R -e "IRkernel::installspec()"
    

For more information, see the IRKernel documentation.

References

  1. http://jupyter.org/about.html
  2. http://www.jupyter.org/