Resources

Bravo cluster - 8 nodes, 8 cores each, 192GB RAM, 10G IB
Run 4 to 8 VMs per node
 
India cluster - 128 nodes. Can run Eucalyptus.


 

Stampede/Pegasus VM's on India FG Cluster

India cluster - 128 nodes. Can run Eucalyptus.


Relevant FutureGrid Links

 

Getting Started with FutureGrid and Applying for Accounts - FutureGrid

Saving an Eucalytpus VM -  http://www.isi.edu/~voeckler/braindump/eucaos-save-active-vm.html

Creating a movable Stampede VM - http://www.isi.edu/~voeckler/braindump/centos-create-movable-virtual-machine.html

Stampede VM's on India

 

The stampede VM's are configured to talk to each other. There is a manual step of updating a condor config file on obelix that the worker node vm's pull when start up. This way they know how to what submit host VM to talk to .

The condor config file is hosted at

http://obelix.isi.edu/stampede-workflows/futuregrid/condor-reconfig.sh

Submit Host VM

$ euca-describe-images | grep stampede/pegasus-sh
IMAGE emi-C18E103B stampede/pegasus-sh.x64.manifest.xml vahi available public x86_64 machine eki-78EF12D2 eri-5BB61255 instance-store

Description

  1. Condor 7.6.6 installed and configured to run as submit host
  2. Pegasus 4.0.0-cvs installed
  3. Latest Periodograms Code installed in /usr/local/periodograms
  4. stampede montage bundle in /home/tutorial/stampede-montage-fg-021512.tgz
$ euca-run-instances -k userkey -n 1 emi-C18E103B -t c1.medium

Worker Node VM

$ euca-describe-images | grep stampede/pegasus-sh
IMAGE emi-186911ED stampede/pegasus-worker.x64.manifest.xml vahi available public x86_64 machine eki-78EF12D2 eri-5BB61255 instance-store

 Description

  1. Condor 7.6.6 installed and configured to run as worker node
  2. Pegasus 4.0.0-cvs as a worker install.
  3. Latest Periodograms Code installed in /usr/local/periodograms
$ euca-run-instances -k userkey -n 1 emi-186911ED -t c1.medium

Sample Steps to start the Stampede VM's

 

Step 1
$ euca-run-instances -k userkey -n 1 emi-CXXXXX -t c1.medium
# wait for the VM to start