Jupyter Notebook on LCC

Jupyter Notebook allows users to run interactive Python notebooks on the LCC cluster. There are several ways to launch a notebook depending on your workflow and software requirements.

Option A – Use Miniconda/conda

Users can create a Conda environment on LCC and launch Jupyter Notebook from that environment.

Typical workflow:

1. Load Conda or activate your Conda environment.

2. Install Jupyter if it is not already installed.

3. Start a notebook server.

Example command:

You can start that notebook on a compute node and access it through a browser interactively using the directions at Open OnDemand OOD Web Portal for CCS.

Option B – Use Jupyter container examples

CCS provides Jupyter containers (for CPU and GPU).

The job submission scripts are in the folder: /share/examples/LCC/jupyter

After you submit the job using the job submission script (submit.sh), you will receive an automated email once the notebook is running on a cluster compute node. The email will include instructions on how to access the notebook, including the compute node and port details.

Option C – Use Open OnDemand with an interactive VNC session

1. Open an interactive session using the directions at Open OnDemand OOD Web Portal for CCS.

2. Once your VNC session is active, open a terminal in your session.

3. Go to the folder where you want to start the Jupyter Notebook.

4. Run the following commands:

   module load ccs/conda/notebook-6.1.4 start-notebook

Open

Screenshot of a running Jupyter Notebook server within VNC session.

5. After your notebook session starts, select the ccs-software logo at the bottom of the VNC session, open the Firefox browser, and type the URL shown in the terminal to open the notebook.

   Open

   

   Screenshot of a running Jupyter Notebook in the browser within a VNC session.

Additional Notes

CCS supports Jupyter Notebooks on the HPC clusters. If you need an account to run jupyter notebooks, please submit a ticket for a new account though the CCS Service Desk.