Skip to content

OnDemand

Login into OnDemand

Login to OnDemand at ondemand.cs.duke.edu.

The OnDemand Dashboard will then open. From there, you can use the menus across the top of the page to manage files, get a shell on the login nodes, submit jobs or open interactive applications.

To end your OnDemand session, click on the Log Out link at the top right of the Dashboard window and close your browser.

Troubleshooting Connection

If you’re encountering issues with the OnDemand web interface, try clearing your browser cache. This can often resolve issues with outdated or corrupted cached files Clearing Your Browser Cache

Managing your files

To create, edit or move files, click on the Files menu from the Dashboard page. A dropdown menu will appear, listing your home directory and the work directory on the CSC. Choosing one of the file spaces opens the File Explorer in a new browser tab. The files in the selected directory are listed.

You can navigate to any directory mounted on the CSC by manually entering the full file system path.

Once you are in the correct directory, you can transfer files to and from the CSC using the CSC OnDemand web interface.

Working with Jobs

You can view current CSC jobs, create new job scripts, edit existing scripts, and submit them to the scheduler through the CSC OnDemand interface.

To access the job management tools, use the “Jobs” menu. For more information about using the tools, please visit the OSC job management page.

Getting a shell

You can get shell access to the CSC by choosing Clusters > CSC Shell Access from the top menu in the OnDemand Dashboard.

In the window that will open, you’ll be logged in to one of the CSC’s login nodes, exactly as if you were using SSH to connect. Except you don’t need to install any SSH client on your local machine.

Interactive applications

One of the main features of CSC OnDemand is the ability to run interactive applications directly from the web interface, without leaving your web browser.

Altering the jupyter environment

The default applications should have an appropriate environment to run applications on the cluster. If you would like to use your own environment (e.g. anaconda) you can change the defaults from:

# Restore module environment to avoid conflicts
module restore
# Load required modules
module load miniconda/23.9.0

to something like:

source /usr/xtmp/NETID/miniconda3/etc/profile.d/conda.sh
export PATH="/usr/xtmp/NETID/miniconda3/bin:$PATH"
conda activate *MYENVIRONMENT*