Open OnDemand provides a web interface to interact with HSUper using any standard browser. It eliminates the need for local SSH clients and offers a graphical dashboard for foundational cluster workflows out of the box.
From the web interface, you can immediately utilize the browser-based File Manager to upload, download, and edit code via a built-in text editor with syntax highlighting. For command-line tasks, an integrated terminal launches native SSH sessions (to login nodes), while a dedicated job management interface allows you to compose, submit, and track the status of your Slurm batch jobs.
Launching interactive graphical applications like JupyterLab or virtual desktops requires specific cluster partition allocations and may require loading application modules on a terminal after being connected to a “HSUper Desktop”.
Contact support if you have any questions or concerns not addressed here.
Navigate to hsuper.hsu-hh.de using your preferred web browser. Log in with your RZ credentials to proceed.
Only users that were authorized to utilize HSUper computing resources are able to login.
To access your data, click the “Files” dropdown menu in the main navigation bar at the top of the Open OnDemand web interface. From the resulting menu, select your “Home Directory” to launch the web-based File Manager. Once open, this interface provides an intuitive graphical view of your remote cluster storage, allowing you to seamlessly drag and drop files to upload them from your local machine, download results, copy or move directories, and launch an integrated file editor directly inside your browser tab without requiring an external FTP client.
While browsing a directory, you can click the “Open in Terminal” button located in the top-right toolbar to instantly open a new shell session pre-navigated to that exact folder path. This terminal utility is highly useful for executing command-line tasks on the cluster file system such as manually running unzip or tar -xvf to extract uploaded datasets.
To manage your cluster compute tasks, click the “Jobs” dropdown menu in the top navigation bar of the Open OnDemand web interface. From this menu, you can select “Active Jobs” or “Job Composer”:
Select this to open a live graphical monitor of the Slurm queue. This interface displays an overview of currently running and pending jobs, enhanced with details such as the assigned Job ID, requested nodes, time elapsed, and the current queue status (such as Running or Queued).
By default, the interface may show all cluster jobs. Use the filter settings (e.g., “Your Jobs”) to display only your own workloads. If a job is stuck or no longer needed, you can delete it directly from this panel with a single click, eliminating the need to type
scancelcommands in a terminal.
For creating and submitting new cluster workflows, the dashboard also includes a “Job Composer” tool accessible under the same menu. This assistant allows you to write, edit, and organize your Slurm submit scripts entirely in your web browser. You can select from pre-configured job templates, easily modify job parameters like walltime or partition settings via standard text fields, and hit submit to queue your job instantly.
To open a shell in your web browser, click on the “Clusters” dropdown menu in the main navigation bar and select “HSUper Shell Access”. This launches a browser-based terminal session on a login node with your home folder as a starting point.
Alternatively, while using the File Manager, you can click the “Open in Terminal” button at any time to instantly open a new shell session pre-navigated to that exact folder path.
The main navigation bar at the top of the page features an “Interactive Apps” dropdown menu. Selecting this option displays graphical applications that run directly on the cluster:
This application provisions a full, interactive graphical Linux desktop environment on a compute node via a Slurm allocation. You will receive exclusive access to the number of CPU cores configured in your submission form. This utility functions identically to an interactive batch job, making it ideal for running GUI-based software packages without occupying login nodes.
To use preinstalled software packages you might need to open a terminal to load the corresponding module and start the application.
This application launches a JupyterLab server on a compute node. The submission form allows you to customize your resources by requesting a specific number of CPU cores and/or GPUs.
If your workflow requires GPU acceleration, you must explicitly select a Slurm partition containing GPU nodes from the dropdown menu in the form and the number of GPUs needed.
“JupyterLab enables you to work with documents and activities such as Jupyter notebooks, text editors, terminals, and custom components in a flexible, integrated, and extensible manner.” Source: JupyterLab Documentation
Click the “My Interactive Sessions” tab in the top navigation bar to access an overview and control center for your running graphical environments (such as HSUper Desktops or JupyterLab sessions). For each running session, you can view the remaining allocation time, check the assigned compute node, adjust the video connection quality (for HSUper Desktops), or instantly delete the session to free up cluster resources. If an application is ready, this is also where you click the launch or connect button to open the graphical interface in a new browser tab.
If you accidentally close your browser tab or experience a temporary network disconnection, your session will continue running on the compute node. Simply return to this page and click the “Connect to JypiterLab” or “Launch HSUper Desktop” button again to resume your work exactly where you left off.
For comprehensive configuration schemas, structural requirements, and step-by-step app creation tutorials, refer to the Open OnDemand v3.1 Interactive App Development Guide.