Connecting via Open OnDemand¶

Overview¶

Open OnDemand is a project from Ohio State University which provides a user-friendly interface for an HPC cluster. You can upload and download files, create and submit batch jobs, view the cluster state, and launch applications that create interactive sessions on compute nodes.

For more info on interactive sessions, please see the Open OnDemand Desktop section.

Support¶

If you have any issues that occur with OnDemand, please email us. Be sure to include the following information:

What you were doing (Interactive Session, Application, Job Composer, etc)
If applicable, the session ID of the interactive session

You can find the session IDs by clicking the My Interactive Sessions button on the top of the page.

Connecting from Campus¶

If you are connecting from any of the University of Kansas campuses, you should be able to reach Open OnDemand in your web browser. Log in with your KU credentials.

Connecting from Off-Campus¶

If you wish to connect to the I2S Cluster from off campus, you must connect to KU Anywhere (VPN) first. More information is available at KU Anywhere. You will still use your KU login to access the cluster when using KU Anywhere.

OnDemand Applications¶

OnDemand Desktop¶

This is the recommended way of getting GUI access to the I2S cluster. Go to the OnDemand Portal in your web browser, and sign in with your KU account. Click on I2S Cluster Desktop and update the page to reflect the resources you need. The Account field is not active as of now, so you can leave that blank. The Partition field defaults to the intel partition.

The user can optionally set the gres flag to make a GPU available, if the partition is set to gpu. See Available GRES devices on I2S Cluster for a list of available GPUs. For example, if one NVIDIA V100 was needed, you would put gpu:v100:1 in the gres field.

Once you are satisfied, hit the "Launch" button to queue your request. You can monitor the state of your request under the My Interactive Sessions tab. Once your session is running, you can click Launch I2S Cluster Desktop to view an XFCE desktop streamed to your browser from the compute node. From here, you can open the terminal and run your application(s).

Jupyter¶

The Jupyter application allows you to create a wide variety of different enviroments in which to work in. Most of the options contained within this app are self-explanatory, but there are a few that require more explanation.

Python Version¶

The python version chosen here will be used to create a conda environment of that version.

Jupyter Variant¶

We support both Jupyter Lab and Jupyter Notebook, but we would recommend most users stick to Jupyter Lab.

Conda Packages¶

This is the main thing you will be changing in order to use different software configurations. For example, if you wanted to create an environment with numpy and Tensorflow, you would enter the following:

tensorflow,numpy

Note

Tensorflow only supports Python 3.9 - 3.12

Any package you enter here must be within the main anaconda repository, conda-forge, or r. To find the correct package names and check if they are included, please go to https://anaconda.org/. If there is a repository you would want added, please let us know via email!

MATLAB¶

The MATLAB application will automatically launch MATLAB in a minimal desktop environment on a cluster node.

Code Server¶

Code Server launches a VSCode-like interface in your browser from a cluster node. This application avoids VNC entirely, minimizing latency.

I2S Cluster Shell Access¶

This is useful if you need to quickly get a command-line interface on one of the front nodes from your browser. The connection will time out quickly, so we do not recommend using this for anything outside of issuing a few quick commands.

OnDemand Files Tab¶

You can use this to manage all your files on the cluster.

You can upload/download files to the cluster, move/copy files, and even edit files. By default, it will land you in your $HOME directory. To navigate to your $SCRATCH or $WORK folders, click on the first / in the path to your $HOME, or use symlinks.

OnDemand Jobs Tab¶

By default, this only shows your jobs running on the cluster. To see all jobs, click on Your Jobs and toggle it to All Jobs.

You can expand jobs to see more information about your job, including requested resources and job time limits. To cancel any job belonging to you, you can click the Delete Job (Trash can icon) button on the right side of the job column.

Firefox¶

If you are needing a web browser for Jupyter Lab/Notebooks, you can load the Firefox module using our modules system. Once the Firefox module is loaded, you can launch Firefox with firefox in a terminal window.