Interactive Apps

Last updated July 05, 2023

We currently offer JupyterLab, RStudio Server, MATLAB, Mathematica, COMSOL and Stata access on both the Discovery and Endeavour clusters.

0.0.1 Starting an interactive session

To start an interactive session in OnDemand, select the application you want to use from the Interactive App dropdown menu at the top of the OnDemand dashboard.

Once you select an application, the session creation page will appear.

Customize your interactive session using the following form fields:

  • Cluster: Select either the general-use Discovery cluster or the Endeavour condo cluster. You must belong to a project with access to the Endeavour cluster in order to select this cluster; otherwise, you will receive an sbatch error (Invalid account or account/partition combination specified).
  • Version (MATLAB and RStudio only): Select from the available versions in the dropdown.
  • Account: The project account to charge resources to. The dropdown should contain all projects you are a member of, and they will be of the form <PI_name>_<project_ID>.
  • Partition: The cluster partition to submit the job to. See the Discovery Resource Overview for more information on the different partitions on Discovery. Depending on the partition you choose, you may not see all of the form fields in this list.
  • Number of CPUs: Number of CPU cores to allocate for your session. The maximum number of cores available depends on the partition selected above.
  • Memory (GB): Amount of memory to allocate. If left blank, a default of 2 GB of memory per CPU core will be allocated.
  • GPU Type (optional): Type of GPU you wish to use. This field is optional and will only be displayed for partitions that have GPU resources. Available GPU types also depend on the partition selected above.
  • Number of GPUs (optional): Number of GPUs to allocate. Maximum of 2. This field is optional and will only be displayed for partitions that have GPU resources. When this field is left empty, Slurm defaults will be applied.
  • Number of hours: Amount of time to allocate to your session. Maximum number of hours varies depending on cluster and partition selected. When the time allocated runs out, your session will end.
  • Constraint: Node features to request (currently, CPU model).
  • BLAS library (COMSOL only): Select from the available libraries in the dropdown. Each library is optimized for a specific CPU architecture.

If you would like to receive an email notification when your session starts, check the corresponding checkbox at the end of the form. Otherwise, click Launch to launch your session:

This will schedule your session in the Slurm queue. You will be taken to the My Interactive Sessions page, where your session will be highlighted in blue with a status of “Starting” or “Queued”:

When your session is ready to use, you will see it highlighted in green with a status of “Running”:

RStudio Server sessions will include a username and password for you to use to log in to RStudio:

MATLAB and Stata include sliders to set Compression and Image Quality:

MATLAB, Mathematica, COMSOL, and Stata run on VNC software.

  • Compression refers to how much the app stream gets compressed. If you have a low-bandwidth network connection, you’ll want to choose a higher compression level for a smoother experience. If you have a slow CPU in your computer, you’ll want to choose a lower compression level.
  • Image Quality refers to the encoder settings used by VNC and will affect the visual quality of the app. Lower quality settings use less bandwidth, so adjust accordingly to your network connection.

Click Connect for Jupyter and RStudio Server or Launch for MATLAB and Stata to open your session in a new browser tab.

0.0.1.1 Managing your interactive sessions

You can view all of your OnDemand interactive sessions by selecting My Interactive Sessions in the main OnDemand menu. Each session will include the following information:

  • Host: Clicking the host will open a shell with access to the node that your session is running on.
  • Created at: The date and time the session was created.
  • Time Remaining: The time remaining for your session, based on the number of hours you specified when creating the session.
  • Session ID: Clicking the session ID will open the file directory for the session, which contains configuration files, logs, and scripts for your app session.
  • Username/Password (RStudio only): The credentials to use when logging in to RStudio Server.

If you no longer need an interactive session, you can delete it using the red Delete button.

0.0.2 Additional resources