Running Notebooks on HPC Systems

This tutorial shows you how to run Jupyter Lab or Notebooks on Comet, TSCC, or TSCC-Stratus using different types of connections to these services, and running them on different nodes (login, interactive, and compute).

Note

It is against SDSC policy to run applications, including notebooks, on the login node. Notebook should be run on compute nodes only.

Once the notebook is running, you can access a the notebook directly from a browser running on your local system (laptop, workstation, etc.).

Note

Unless you are using SSH Tunneling or the Reverse Proxy methods, the connection is insecure and has the potential to be hacked by malicious parties.

On Comet, we only support the following connection scenarios:

• Connection to Notebook over SSH tunneling (secure)
• Connection to Notebook over HTTPS using the Reverse Proxy Service (very secure)

In addition, notebooks can be run on the following nodes:

• Login node
• Interactive node
• Compute node
• GPU node

The choices and combinations used to run notebooks affect the security and efficiency of you application. In the next sections, we’ll describe how to connect and how to run the notebooks and the security impacts of these configurations.

• Jupyter Notebook Overview
• Example Notebooks
  • Beginner Tutorials
  • More Advanced
• Software Prerequisites
  • Install Miniconda
  • Install Jupyter Notebook
  • Install JupyterLab
  • Other Python Packages**
  • Setup a Conda Virtual Environment
  • Other Python Packages
  • Download Example Notebooks
  • Basic HPC Skills
• Insecurity with direct node access
  • Log onto comet.sdsc.edu
  • Launch a notebook on the login node:
  • Obtain an interactive node:
• Security with SSH Tunneling
  • Interactive Node Method
  • Compute Node Method
• Security with Reverse Proxy Service
  • Overview
  • Clone the RPS repository
  • Launching the Notebook
  • Usage
  • Example Commands
  • Example Output
• About the Team
• Contact Us