FAQ
Overview
Who uses Open OnDemand?
Open OnDemand is currently used by thousands of institutions in academia and business in more than 100 countries across the world.
Open OnDemand allows system administrators to quickly and easily connect many researchers to institutional compute resources and provides an accessible platform for researchers to run common software without computing expertise.
Who funds/supports Open OnDemand?
While being open source and community supported, Open OnDemand has also been the recipient of over $10 million of NSF grants that has enabled teams at the Ohio Supercomputer Center (OSC), the Massachusetts Green High Performance Computing Center (MGHPCC), Texas A&M University, and the University of Maryland to dedicate full time staff towards enhancing and maintaining the project.
How secure is Open OnDemand?
Open OnDemand integrates seamlessly with common institutional authorization methods and utilizes the security and permissions capability of the underlying operating system to manage user groups and levels of access. Open OnDemand is trusted and used by U.S. federal agencies and cutting-edge technology enterprises alike, and we regularly test against known attacks, scan dependencies, and release security patches for all supported releases.
How are CVEs detected, handled, and reported to the community?
GitHub automatically scans our codebase and warns about potential risks, which are evaluated by our development team. Based on this evaluation we then make appropriate patches and publish the full report—detailing the issue, severity, versions affected, and what patches were made for which releases in our security advisories.
Resources
How do I start using Open OnDemand at my institution?
We publish and maintain our installation guide, which walks you through supported operating systems, required dependencies, and integration with authenticators and schedulers. Open OnDemand maintains RPM distributions to build the production instance and upgrading, so most of the setup work is integration with the institutional services you already use.
Will Open OnDemand be able to run my custom tools?
Open OnDemand is designed to be a platform for all varieties of HPC applications, seen in the wide variety of applications that users across the community have developed for their systems. Contained in the official Open OnDemand distribution are shell and desktop apps, giving you the same familiar access points to your HPC resources that your users are accustomed to. Many apps are run using Passenger, which supports Ruby on Rails, WSGI, and Node.js apps, in addition to “interactive apps,” which streamline the startup and access of software tools like MATLAB and Jupyter. Detailed instructions on app development—as well as tutorials for developing both Passenger and interactive apps—can be found in the app development how-tos.
How customizable is Open OnDemand?
Open OnDemand is highly customizable, giving you full control over layout and appearance wherever possible. A detailed analysis of customizable features can be found in the customizations overview.
Where do I go for help if I have problems with my Open OnDemand instance?
The first place to bring any issue you encounter with Open OnDemand is our community Discourse, which connects you with a large Open OnDemand community that can give advice and share solutions to many common problems or questions. Our full-time developers are also very active on Discourse, investigating problems and offering individualized solutions and expert advice. For more dedicated support, consider our support plan and find us at conferences like SC, ISC, PEARC, and our very own GOOD conference.
We also offer Open OnDemand Office Hours, a monthly virtual meeting with the Open OnDemand development team to help diagnose and solve persistent issues.
How many people are typically required to maintain an institutional instance? How does this scale?
Institutions typically find that Open OnDemand configuration and maintenance is easily handled by a single system administrator without significant impact on other duties they may have. While the initial setup can require some time and troubleshooting, future updates and continued maintenance are generally light work. Scaling issues could arise if the login node hosting your Open OnDemand instance is overloaded, and may require moving the instance to a new machine or allocating more resources towards it.