Notre Dame Computational Grid

The University of Notre Dame is a member of the Northwest Indiana Computational Grid and a virtual organization (VO) of the Open Science Grid (OSG). The Center for Research Computing maintains a set of resources devoted to high job volume/throughput computing.

Getting Started with Notre Dame NWICG Machines

In order to access the grid resources, you must have a certificate from one of below CAs:

Instructions for obtaining a CERN certificate:

Instructions for obtaining an OSG certificate:


The specific computational hardware at ND for which you will have priority is:

For current information on grid partner's hardware review:

Submitting Jobs to ND NWICG Machines

Staging Files
The head node DOES support the CRC /dscratch and /scratch365 distributed files space.
You can SCP files to your user space on or move them into /dscratch or /scratch365 from AFS on any CRC head node.

OSG utilizes a number of Globus Toolkit components to manage grid security, allocation, and management. The Globus Grid Resource Allocation Manager (GRAM) is the primary tool for job submission. The software is designed for maximum capability over a wide range of computation resources, with that ability however also comes a degree of complexity. The directions which follow are the bare minimum functionality to get you going. Please explore the additional capabilities.

Before submitting each job make sure your grid proxy is current. The following command will validate your grid proxy for 12hrs. Remember to log into for OSG and for NDCMS.

$ grid-proxy-init

Test your authority with the gatekeeper

$ globusrun -a -r

Example production batch job scripts for use with the following submission command. The 'condor' job manager designation instructs the job to be directed to an condor managed pool of resources.

$ globus-job-submit -s
  • Example simple Job Script.
echo "This is a GRAM test with Condor."
  • Example Condor Job Script.
universe = vanilla
executable =
arguments = hello condor
output = simple.stdout
error = simple.stderr
should_transfer_files = yes
when_to_transfer_output = on_exit
log = simple.logfile
queue 2