Execute simple MPI job on Ranger using Apache Airavata

In an earlier post[3] we looked at how to install a simple hello world MPI program in Ranger [1]. In this post we will look at how to execute the previously installed application on Ranger using Apache Airavata.

1. Before starting airavata configure repository.properties file by modifying default fake values for following properties.

trusted.cert.location=path to certificates for ranger
myproxy.server=myproxy.teragrid.org
myproxy.user=username
myproxy.pass=password
myproxy.life=3600

Configure your certificate location and my proxy credentials and start Jackrabbit instance and airavata server with all the services.

2. To create a workflow to run on ranger you need to create Host Description, Application Description and a Service Description. If you  don’t know know how to create them refer airavata 10 minutes article [2] to understand how to create those documents. Once you become familiar with XBaya UI, use following values for fields given below and create documents using XBaya GUI.


Host Description

• Host Address - gatekeeper.ranger.tacc.teragrid.org

Click on the check box - Define this host as Globus host, then following two entries will enable to fillup.

• Globus Gate Keeper Endpoint - gatekeeper.ranger.tacc.teragrid.org:2119/jobmanager-sge  
• Grid FTP Endpoint                   - gsiftp://gridftp.ranger.tacc.teragrid.org:2811/

Service Description

Fill the values similarly like in 10 minutes article [2] service description saving part.



Application Description

• Executable path - /share/home/01437/ogce/airavata-test/mpi-hellow-world
• Temporary Directory - /scratch/01437/ogce/test

Select the above created service descriptor and host Descriptor.. When you select the above created host descriptor you will see a new button Gram Configuration. Now click on that and fill following values.

• Job Type - MPI
• Project Account Number - You will see this when you login to ranger, put your project Account number in this field

• Project Description - Description of the project - not mandetory
• Queue Type - development

Click on Update button and save the Application Description.

Now you have successfully created Descriptors and create a MPI workflow like you have done in 10 minutes article [2] and try to run it.



[1] - http://www.tacc.utexas.edu/user-services/user-guides/ranger-user-guide
[2] - http://incubator.apache.org/airavata/documentation/system/airavata-in-10-minutes.html
[3] - http://heshans.blogspot.com/2012/06/running-simple-mpi-job-on-ranger.html