Skip to main content

Interactive Apps — Interactive Sessions

The Interactive Apps section of Open OnDemand allows you to launch graphical or interactive applications directly on MCMeSU compute nodes, accessible through your browser.

Interactive sessions give you a real-time interface with the compute resources — ideal for data exploration, visualization, iterative development, or any workflow that requires immediate feedback.

tip

Interactive Apps run as regular Slurm jobs under the hood. They consume resources from the same partitions as your batch jobs, and count against the same allocations.


How It Works​

When you launch an Interactive App:

  1. Open OnDemand submits a Slurm job on your behalf to the requested partition
  2. The job starts on a compute node and launches the application
  3. OOD establishes a secure connection between your browser and the running application
  4. You interact with the application directly in your browser tab
  5. When you are done, terminating the session cancels the underlying Slurm job and frees the resources

Available Applications​

ApplicationDescription
Jupyter NotebookInteractive Python, R and Julia computing environment
RStudio ServerR environment for statistical computing and bioinformatics
Code ServerVisual Studio Code in the browser
Remote 3D DesktopVisualize your results in 3D
ParaviewPost-processing visualization engine
Visual Molecular Dynamics (VMD)Molecular visualization program for displaying, animating, and analyzing large biomolecular systems using 3-D graphics and built-in scripting
note

Additional interactive applications may be added in the future. Contact the MCMeSU team if you need a specific application deployed on the portal.


Application Forms​

Each Interactive App has its own resource request form. Fill it in carefully — you cannot change these values once the session has started.

FieldDescription
Extra Jupyter ArgsSpecifies the extra jupyter arguments
PartitionTarget Slurm partition (bigm, std, thin or gpu)
Job's timeWall time limit for the session (in hours, limited to 24 hours, contact MCMeSU team if you need more time)
Number of nodesNodes available to your notebook (limited to 1, contact MCMeSU team if you need more)
Number of CPUsCPU cores available to your notebook
Memory (GB)RAM available to your notebook
Number of GPUsNumber of GPUs requested (GPU partition only)
Be notified by emailCheck this box to be notify by email when the session starts and ends

Launching an Interactive App​

interactive-apps-001

  1. Click Interactive Apps in the top navigation bar
  2. Select the application you want to launch from the dropdown menu
  3. Fill in the resource request form — this is specific to each application (see the dedicated page for details)
  4. Click Launch
  5. You are redirected to the My Interactive Sessions page, where your session will appear with a Queued status

My Interactive Sessions​

The My Interactive Sessions page is your control panel for all active and recently completed interactive sessions. It is accessible at any time via Interactive Apps → My Interactive Sessions in the navigation bar.

Session States​

StateMeaning
QueuedThe underlying Slurm job is pending in the queue
StartingThe job has started and the application is initializing
RunningThe session is ready — the Connect button is available
CompletedThe session has ended (time limit reached or manually terminated)
FailedThe session encountered an error during startup

Connecting to a Running Session​

Once your session reaches the Running state:

  1. Click the Connect button on the session card
  2. The application opens in a new browser tab
tip

If you accidentally close the browser tab, you can reconnect to a running session at any time from the My Interactive Sessions page — as long as the underlying Slurm job is still running.

interactive-apps-002

Terminating a Session​

To end an interactive session before its time limit:

  1. Go to My Interactive Sessions
  2. Click the Delete button on the session card
  3. Confirm the termination

This immediately cancels the Slurm job and releases the allocated resources. Always terminate your sessions when you are done — idle sessions continue to consume resources until their time limit is reached.

warning

Do not rely on the time limit to clean up your sessions. An idle Jupyter Notebook left running for hours blocks compute resources that other users could be using.


Best Practices​

  • Request only what you need: overestimating CPUs, memory, or GPUs delays your session start and wastes shared resources.
  • Always terminate idle sessions: if you are done working, click Delete — do not just close the browser tab.
  • Save your work regularly: if the session times out or the node fails, unsaved work in the application will be lost.