Australian Research Environment (ARE)

ARE is a web-based graphical interface for performing your computational research, provided by NCI.
ARE can give you access to NCI’s Gadi supercomputer and data collections.

There are multiple applications included in ARE, but the two most used for ACCESS-related activities are Virtual Desktop (VDI) and JupyterLab.

Prerequisites

To use ARE, you must have an NCI account and be a member of a project with computing resources (SU). If you are new to ACCESS, follow the First Steps.

Start an ARE session

Virtual Desktop (VDI) JupyterLab

To start an ARE VDI session go to the ARE VDI Desktop page.

To start an ARE JupyterLab session go to the ARE JupyterLab page.

Session options

Launching an ARE session is similar to submitting an interactive PBS job that enables you to connect to a Gadi computing node.
Therefore, there are multiple PBS directives and other options you can select:

• Walltime (hours)
  Number of hours your VDI session will run for (unless manually ended earlier).
  The maximum number of hours an ARE session can run for depends on the selected Queue. For more information, check Gadi Queue Limits.
  Once the session ends any operation still in progress on the session's computing node(s) will be immediately terminated.
• Queue
  Gadi queue that your session will be scheduled in. For more information check Gadi Queue Structure.
• Compute Size
  Amount of resources (CPUs, Memory, etc.) available to your session.
  You can either choose a pre-configured option, or select a custom one (e.g., cpus=6 mem=40G).
• Project
  The NCI project the ARE session will be charged to.
  You must be a member of the specified project.
  The specified project must have allocated Service Units (SU).
  For more information, check how to join relevant NCI projects.
• Storage
  /g/data (inserted as gdata/<project-ID>) and /scratch (inserted as scratch/<project-ID>) data storage projects that will be available to the session.
  In ARE, data storage locations need to be explicitly defined. This means that you need to insert any /g/data and /scratch project folders you want to execute data I/O operations from.
  Multiple storage projects are separated by a plus (+) (e.g., gdata/tm70+gdata/hh5+scratch/xp65).
  Generally, you need to be a member of the specified projects to access their storage data.
  If you try to access data from a project folder not included in the session's storage, you will get an error similar to the following:
  <cmd>: cannot access '</path/to/file>': No such file or directory
• Software
  Software licenses that are requested for your ARE session. Leave blank if no license needed.
  Multiple licenses are separated by a colon (:).

Advanced options

Extra arguments
Additional arguments to pass on the JupyterLab command line (e.g., --debug, --log-level=INFO)
Module directories
Include module directories.
It is the eqivalent of module use </path/to/module/directory> run on the command line.
You also need to include the project directory of each module directory in the Storage option.
Modules
Include modules. It is the equivalent of module load <module-name> run on the command line.
If the module is not inside Gadi's default module directory /apps/Modules/modulefiles, you need to include the module directory in the Module directories option.
Python or Conda virtual environment base
Path to a Python or conda base environment to be activated for the JupyterLab session.
It is the equivalent of source <path/to/environment/bin/activate> run on the command line.
You also need to include the project directory of the virtual environment in the Storage option.
Conda environment
Name of a specific conda environment to be activated for the JupyterLab session.
It is the equivalent of conda activate <environment-name> run on the command line.
You need to include the path to the conda base environment in the Python or Conda virtual environment base option.
• Environment variables
  Environment variables passed to the session. Identical to the -v PBS directive.
  Multiple environment variables are separated by a comma (,).
• Jobfs size
  The maximum amount of local disk available to the session.
  If missing, it is automatically set to 100MB.
• PBS flags
  Extra PBS directives to be used for the ARE job submission.
• Pre-script
  A script or executable that will be run just before starting the ARE session.

Launch the session

1. Click on the Launch button to launch the session. You will be prompted to your Interactive Sessions page and you will see your last requested session at the top.
2. Wait until your session starts and then click on the Launch VDI Desktop button to open a new tab with the VDI interface.
   Inside the VDI interface, you can open the terminal by clicking on the black terminal icon at the top of the window.

   Wait until your session starts and then click on the Open JupyterLab button to open a new tab with the JupyterLab interface.
   Inside the JupyterLab interface, you can open a new notebook by clicking on the Python3 Notebook button in the Launcher panel (to open a new Laucher panel, click on the plus button next to your current tab).

Delete an ARE session

You can delete a session before its automatic expiration (based on the specified Walltime) by clicking on the session's Delete button in the Interactive Sessions page.

References

• https://opus.nci.org.au/display/Help/ARE+User+Guide
• https://are.nci.org.au/pun/sys/dashboard/batch_connect/sys/desktop_vnc/ncigadi/session_contexts/new
• https://are.nci.org.au/pun/sys/dashboard/batch_connect/sys/jupyter/ncigadi/session_contexts/new

---

Last update: April 25, 2024