Skip to main content

Open OnDemand

Open OnDemand (OOD) is a web-based portal that provides an integrated, single access point for remote HPC Cluster.

It enables you to perform many common tasks without CLI, including:

  • Managing files (upload, download, edit)
  • Monitoring running jobs and resource usage
  • Launching interactive jobs (e.g., Jupyter, RStudio, VS Code)
  • Displaying remote desktop for software graphical interface (e.g., MATLAB, COMSOL)
info

UW VPN is required to access OOD if you are off-campus. You can find instructions on how to connect to the UW VPN here.

JupyterLab and Jupyter Notebook​

You can run JupyterLab (recommended) or the classic Jupyter Notebook interface through OOD's Interactive Apps menu.

Launch Jupyter​

Step 1: Navigate to the Jupyter App

From the OOD dashboard, select Interactive Apps > Jupyter.

The default interface is JupyterLab. If you prefer the classic interface, choose “Jupyter Notebook” instead.

Step 2: Launch An Interactive Job

You will be prompted to enter information about the job you want to run.

Once you've filled out the form, click Launch.

Step 3: Wait for Allocation

After submission, OOD will:

  1. Submit a Slurm interactive job under your account.
  2. Wait for resources to become available.
  3. When ready, show a Connect to Jupyter button.

Click Connect to Jupyter once it appears. A new browser tab will open with your live Jupyter environment running on a Tillicum compute node.

Use Conda Environments as Python Kernels​

By default, Jupyter uses the system's Python kernel provided by a minimal Python environment. There are also several prebuilt Python kernels with more packages installed.

You can also use your own Conda environments as custom Python kernels.

Step 1: Load Conda and Activate Your Environment

If you already have a conda environment you want to use as a Jupyter kernel, make sure it includes the IPython kernel package ipykernel. Otherwise, create a new Conda environment with the packages you want plus ipykernel.

From a terminal (either SSH or OOD's "Tillicum Login"), run:

module load conda
conda activate myenv
conda install ipykernel

Step 2: Register Your Environment as a Jupyter Kernel

Run ipykernel install in your activated environment to set up a Jupyter kernelspec.

python -m ipykernel install --user --name myenv --display-name "Python (myenv)"

This creates a Jupyter kernel entry visible in Jupyter notebook. Any packages installed in your Conda environment will automatically be available to you. Please note:

  • --name: internal environment name
  • --display-name: name shown in JupyterLab
tip

Containers can also be registered as Jupyter kernels.

Step 3: Launch A Jupyter Notebook

Once JupyterLab opens, in the Launcher tab, under Notebook choose Python (myenv).

Your notebook will now run inside your custom Conda environment.

Step 4: Switch the Kernel

Inside a notebook, go to Kernel > Change Kernel to switch kernels in the dropdown box.

💡 TIP: If your custom kernel doesn't appear, restart the Jupyter session after running the ipykernel install command.

Close the Session​

When you finish:

  1. Save your work and close the Jupyter browser tab.
  2. Return to OOD and click Delete on your running session card.
  3. The compute resources will be released back to the cluster.
warning

Leaving sessions running consumes GPU hours and counts toward your project usage.