.. _s-ga8: ********************************************************************** GA8: Running Condor Tasks on a Grid Appliance ********************************************************************** .. sidebar:: Page Contents .. contents:: :local: Summary: ~~~~~~~~ This tutorial provides step-by-step instructions on how to deploy a Grid Appliance configured with Condor, an open-source job scheduler. The Grid Appliance can be used as a basis for running a variety of educational, training and outreach activities. Prerequisites: ~~~~~~~~~~~~~~ - To follow this tutorial, it is a pre-requisite that you have gone through one of the core Grid appliance tutorials: - :ref:`Running the Grid Appliance on your desktop ` - :ref:`Running the Grid Appliance on FutureGrid ` Set Up and Submit a Condor Job ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ - Once you are logged in through ssh into the Grid appliance, change your user to "griduser", change your directory, and compile a simple program which estimates the value of "pi" using a Monte Carlo method: .. code:: rteindent2 su griduser cd cd examples/montepi gcc montepi.c -o montepi -lm -m32 - Inspect the "submit\_montepi\_vanilla" file - this file sets Condor up to submit two batches of 50 tasks, each of wihch runs one Monte Carlo task. Chech the status of the Condor pool (which shows nodes are connected, available for running tasks, busy, etc) with the "condor\_status" command. Then use the "submit\_montepi\_vanilla" file to submit your job: .. code:: rteindent2 more submit_montepi_vanilla condor_status condor_submit submit_montepi_vanilla - Monitor the progress of your jobs with the "condor\_q" command, which shows the status of your job queue. You can also check the status of the Condor pool with "condor\_status" to see which resources become busy. Once the Monte Carlo tasks complete and there are no jobs in the queue, check the various results that have been obtained from the different tasks with the "grep" command: .. code:: rteindent2 condor_q grep -i pi *.out