Open OnDemand

Here is some getting started information for using Open OnDemand to access HPG.

The OnDemand service is accessible only within the UF campus network and using the campus VPN.

If you are working off campus, you will need to use the VPN to connect to the UF campus network using AnyConnect VPN or similar software. Please see: https://vpn.ufl.edu for additional information.

Open OnDemand

Open OnDemand Open OnDemand is setup to provide a web-only interface to starting and attaching to HiPerGator and it supports GUI programs, Job Status, and other interactions with HPG via seamless connections through web browsers. This service is available at the following URL:

https://ondemand.rc.ufl.edu

To provide feedback, please open a support request with [OnDemand] in the subject. Please note:

Ondemand is currently a Beta service. Users are welcome to test the service and submit support requests with feedback. However, no SLA is guaranteed and all features should be considered experimental.

Getting started video

Please watch our introductory video here: Using Open on Demand on HiPerGator [5 min, 11 sec] This video covers the use of Open on Demand https://openondemand.org/ on HiPerGator via https://ood.rc.ufl.edu/.

• Connect to https://ood.rc.ufl.edu/
• Launch a terminal
• Launch many graphical applications like Rstudio, Matlab and Freeview
• Manage job submission
• Upload, download and edit files via your browser

Navigating OnDemand

To Navigate the basic functions of OnDemand, first login to https://ood.rc.ufl.edu

Once you access the dashboard, you can use the top menus to navigate.

Files

This menu, if selected, will allow you to access your group's storage areas including:

• Home
• Orange
• UFRC

Jobs

This menu shows both the job status and job composer links

• Job Status: This link shows the Job status window. This query's SLRUM and shows you the current status of your pending, running, and recently completed jobs.
• Job Composer: This link allows access to the job composer if you would like to use OnDemand to run your existing SLRUM scripts.

Clusters

• HiPerGator Shell Access: This link opens a web terminal in your browser that allows access the HiPerGator login shell

Interactive Apps

Clicking this menu will present a drop-down menu with a list of the current interactive GUI applications available within OnDemand.

If you don't see an application you need, feel free to select the "Console" application and use the HPG modules system to load and run your application.

My Interactive Sessions

Clicking this link will take to you a list of your currently running interactive OnDemand jobs.

Help

This link provides access to some quick links for help on HPG and for OnDemand.

Logout

This link logs you out of the OnDemand page.

Starting an Interactive application

To start an interactive application, first select the application you would like to start from the "Interactive Applications menu". Then you will see a form with the SLURM options available for that application. Fill in the fields with the resources you would like allocated for this application session.

General Form Fields

Version: Select the application version This defines the version of the application you want to load.

• Number of MPI Tasks (--ntasks): recommended default, 1

• Number of CPU cores requested per MPI task (--cpus-per-task, -p): recommended default, 1

• Maximum memory requested for this job in Gigabytes (--mem, -m): recommended default, 1

Note: If you are using advanced memory options in Additional SLURM Options below, then leave this blank.

• SLURM Account (--account, -A): recommended default, (leave blank)

• QoS (Required if custom Account is set, --qos, -q): recommended default, (leave blank)

• Time Requested for this job in hours (--time, -t): recommended default, 1

• Cluster partition (--partition, -p): Recommended default, (default)

• Generic Resource Request (--gres): Recommended default, (leave blank)

Additional SLURM Options: Recommended default, (leave blank)

You can have multiple copies of the same application running at the same time by submitting the form multiple times.