How to Run Jupyter Notebook¶

We recommend using JupyterLab on Open OnDemand. If that isn't suitable, you can follow this guide.

In this guide we will be discussing how to run Jupyter notebook. You can learn more about it here: https://jupyter.org/

Prerequisites¶

If you are on Mac/Linux, you will just need to use the terminal. For Windows, please use powershell.

Connecting to Pronto when using Jupyter¶

Using Jupyter requires you to connect to an interactive session on Pronto with port forwarding enabled.

Using the terminal, ssh to pronto using:

ssh <YourNetID>@pronto.las.iastate.edu

Once you are on pronto, salloc node using: (adjust the time allocated if desired)

salloc --time=00:10:00 --nodes=1

If you are not familiar with allocating nodes or want to specify other configurations, please refer to this guide.

Take note of the node you were assigned. In the screenshot below, the node I was assigned was "biocrunch".

juptyer_salloc

Open up another terminal and ssh to the assigned node. Here you will need to specify a "-L" flag and port numbers. The last 4 digits will be the port number Jupyter will be listening on, so you may need to change it to your needs. If you choose to omit a port number, Jupyter will pick one for you.

ssh <YourNetID>@<NodeName>.las.iastate.edu -L <SpecifyPortNumber>:127.0.0.1:<SpecifyPortNumber>

In my case, I will choose port 9999. My command would look like

ssh jones@biocrunch.las.iastate.edu -L 9999:127.0.0.1:9999

(After you connect to your specified node, you may get a benign warning message saying "setsockopt IPV6_TCLASS 16: Operation not permitted:". This should not cause any issues.)

Installing Jupyter Notebook¶

Follow our guide to create a virtual environment where jupyter notebook will be installed.

Once you have created and activated the virtual environment, run the following to install Jupyter notebook.

pip install jupyterlab

If you need to use any additional python packages with jupyter, you can install them the same way.

Running Jupyter Notebook¶

Activate the Virtual Environment¶

If you just installed Jupyter notebook as described above, this part is already done.

If you installed Jupyter notebook previously and want to re-use it with the same Python packages as before, you need to activate the virtual environment you installed it in.

For example:

module load python/3.8.0-k7w5uj4  
source /work/jones-lab/jupyter-virtual-environment

Be sure to use the same python module that you used to create the environment. And replace /work/jones-lab ... with the actual path to the directory you created the virtual environment in.

Start Jupyter Notebook¶

Run:

jupyter lab --no-browser --port 9999

Next, open up a browser, and copy the link that was in the terminal and paste it in the browser.

jupyter_address

Your browser should look like this:

jupyter_browser

Running Python on Jupyter¶

You can upload your own python code or other files using the "Upload" button on the left hand side. You also have the option of writing and running python code within the browser itself. To create a new python script, click on the Python icon under the Notebook heading_._

jupyter_new_upload

You will be redirected here:

jupyter_python1

Now, type in the code that you wish to run. In my example, I will just do a simple hello world program. Once you have typed in your code, you can run the program simply by clicking the "run cell" icon (boxed in red below).

run_python

Troubleshooting¶

If you attempt to run jupyter-notebook and get a "Permission denied: /run/user/#####", run:

unset XDG_RUNTIME_DIR

and try again.

Useful links¶

This is only a brief tutorial on how to get Jupyter running on pronto, but you can find more extensive documentation here: https://jupyter.org/documentation

Your session may disconnect if you leave it idling for too long. Please refer to this documentation on how to configure the idle time of your server.

You now know how to run Jupyter notebook!