Synopsis

Besides accessing Snellius via a terminal, the system can also be used via a web browser using the OpenOnDemand demand service. The service allows users to:

  • manage files
  • work in a shell without using a terminal client on their remote system
  • launch jobs based on predefined templates
  • run interactive web services such as RStudio and Jupyter Notebooks
  • run applications with graphical user interfaces  



Access the OpenOnDemand service.

The Snellius OpenOnDemand service can be accessed via the following URL:

https://ondemand.snellius.surf.nl/

To log in, a valid Snellius login is required, and the Usage Agreement needs to be accepted, see Snellius FAQ.  

Overview

The key features are available in the top menu on the start page:

  • File explorer to upload files to / download from the home directory. Files can be deleted, renamed, moved to a different director or edited online using the integrated online editor
  • Job manager to compose and submit Slurm jobs to the cluster
  • Shell access 
  • Interactive applications 
  • Overview of recent interactive sessions

The file explorer

The file manager can be selected in the top menu. By default, the content of the home directory is shown and can be accessed via a web browser. Other file systems (e.g. a project space or the scratch space) can be chosen under “Change directory”.

Files can be edited using the built-in editor available on the context menu of a file.


Shell access

OpenOnDemand allows users to launch a terminal session on the system without using an SSH client. 

To copy and paste into the OPEN OnDemand terminal, use the keyboard shortcuts CTRL+C / CTRL+V

Interactive Sessions

Interactive sessions allow users to execute interactive applications on Snellius. We distinguish two application types:

  1. Server-Based Applications: Some applications are available as web-based or server-hosted versions, which you can access directly through a web browser without needing any local installations. Examples include applications like Jupyter Notebooks, RStudio and VSCode.

  2. Remote Desktop Access: For any other applications with graphical interfaces such as ANSYS, MATLAB, or ParaView that may not have a server-hosted version, you can use Remote Desktop. This option lets you connect to a remote system running the application, providing a fully interactive desktop environment.  

To launch an interactive session, click on one of the interactive applications.



After selecting the application, you need to specify the required resources (time and CPUs / GPUs).

For most applications, default settings are preconfigured to ensure optimal performance and ease of use. These settings include:

Partition: Selected to align with the application’s typical resource requirements.

Wall Time: Set to a standard duration that accommodates common tasks.

Number of CPU Cores: Configured to meet the usual processing needs of the application. 

Memory: Allocated based on expected usage to maintain performance and stability.

GPUs: Number of GPUs required (only valid for GPU partitions)

Reservation: Applied when specific resources need are to be allocated from a preconfigured reservation. 

The default resource allocation for VS Code uses the staging partition with one CPU core. This setup is intended to handle standard code editing and development tasks while minimizing resource usage, which helps to conserve budget by using a lightweight, cost-effective partition.

For users requiring more computational power, an alternative is to use a rome or genoa partition.  However, please note that this option starts with a minimum of 16 cores, which can consume more of your allocated resources and increase costs (see Snellius partitions and accounting).


After clicking the Launch button, a job is submitted to the specified partition. 

When the requested resource is available, the application can be access via the launch button.

An interactive session costs budget (one SBU per CPU core and hour). Therefore, it is recommended to terminate the session when the work is done using the Delete button.

Webserver application

When the resources for a webserver-based application are available, you can use the connect button to open the client in a browser.

Jupyter Notebooks

Once launched, the Jupyter Notebook interface opens in your browser, where you can create and manage notebooks, write code, visualize data, and execute computations directly.



VS Code

In the VS Code web-based interface you can edit code, manage files, install extensions, and run tasks all within the browser, with seamless access to the resources allocated by the server, whether you’re working on a single core in the staging partition or a more powerful setup if needed.

Remote Desktop

The Remote Desktop provides a full graphical environment that you can access from your local device, allowing you to interact with applications hosted on a remote system as if you were working directly on that machine. This is especially useful for applications with complex graphical interfaces or high resource demands that run on a Snellius compute node.

Within the application menu on the Remote Desktop, you’ll find:

  1. Default Applications: These include standard productivity and utility tools commonly available in desktop environments, such as a web browser, file manager, and text editor etc.

  2. Snellius Applications:  Under “Snellius”, you’ll find applications from the Snellius software stack with GUI features such as Ansys, MATLAB, Paraview and TotalView.
  3. The Terminal Emulator gives you command-line access to Snellius, allowing you to manage files, run scripts, and execute commands and load modules from the Snellius software stack.


terminal

Please note that not all software is available directly in the application menu. Some specialized modules or tools may need to be launched from the terminal, which you can open from the application menu.
Using the terminal, you can access the full software stack available on Snellius. This flexibility allows you to work with all installed software, even if it isn’t integrated into the graphical application menu, or it is installed in your local home directory.

On the right you will find an example for launching Gnuplot, demonstrating how to start this software from the command line.



combining applications

In a remote desktop session, it is also possible to work interactively on different applications, which is not the case for server-based applications.