uwsampa/grappa-starcluster

Name: grappa-starcluster

Owner: Sampa

Description: null

Created: 2016-03-25 06:24:18.0

Updated: 2016-06-26 10:25:06.0

Pushed: 2016-05-18 17:11:50.0

Homepage: null

Size: 13

Language: Shell

GitHub Committers

UserMost Recent Commit# Commits

Other Committers

UserEmailMost Recent Commit# Commits

README

grappa-starcluster

Grappa running in the Amazon cloud, using StarCluster.

Setup

Install StarCluster.

Create your config file by running python create_config.py. The config is written to grappacluster.config.

Launch a cluster

Start a new cluster named mygrappacluster. See starcluster for other options.

cluster -c grappacluster.config start -s 2 -i r3.xlarge mygrappacluster

When the cluster has started, ssh into the master.

cluster -c grappacluster.config sshmaster mygrappacluster
Use the cluster

grappa-starcluster uses starcluster's default sgeadmin user for running MPI jobs. We suggest using sgeadmin for everything, other than when you require root.

geadmin

Do this once to create your MPI host file

 'node[0-9]\|master' /etc/hosts >~/hosts

In the home directory, you'll find grappa/ with everything already default configured and installed. If you ever need to re-configure, you can use the command in ~/build_grappa.sh.

Let's run an example job to make sure things work.

/grappa/build/Make+Release
 -j demo-hello_world
un -np 12 --hostfile ~/hosts -- applications/demos/hello_world.exe

If your cluster has 2 nodes then the output should look something like:

8 16:48:48.750394  2201 Grappa.cpp:647]
---------------------
ed memory breakdown:
de total:                   29.7124 GB
cale shared heap total:     14.8562 GB
cale shared heap per core:  2.47604 GB
mmunicator per core:        0.125 GB
sks per core:               0.0156631 GB
obal heap per core:         0.619007 GB
gregator per core:          0.00247955 GB
ared_pool current per core: 4.76837e-07 GB
ared_pool max per core:     0.619009 GB
ee per locale:              10.221 GB
ee per core:                1.7035 GB
---------------------
8 16:48:48.808346  2202 hello_world.cpp:45] Hello world from locale 0 core 1
8 16:48:48.808636  2201 hello_world.cpp:45] Hello world from locale 0 core 0
8 16:48:48.808657  2203 hello_world.cpp:45] Hello world from locale 0 core 2
8 16:48:48.808764  2204 hello_world.cpp:45] Hello world from locale 0 core 3
8 16:48:48.810246  2209 hello_world.cpp:45] Hello world from locale 0 core 5
8 16:48:48.810478  2205 hello_world.cpp:45] Hello world from locale 0 core 4
8 16:48:48.852023  1348 hello_world.cpp:45] Hello world from locale 1 core 7
8 16:48:48.852031  1350 hello_world.cpp:45] Hello world from locale 1 core 8
8 16:48:48.852037  1352 hello_world.cpp:45] Hello world from locale 1 core 9
8 16:48:48.852037  1359 hello_world.cpp:45] Hello world from locale 1 core 11
8 16:48:48.852210  1356 hello_world.cpp:45] Hello world from locale 1 core 10
8 16:48:48.852695  1346 hello_world.cpp:45] Hello world from locale 1 core 6
Rebuilding the AMI

Coming soon…


This work is supported by the National Institutes of Health's National Center for Advancing Translational Sciences, Grant Number U24TR002306. This work is solely the responsibility of the creators and does not necessarily represent the official views of the National Institutes of Health.