You are here: Home / Tools & Utilities / MESA on Clusters / Installing and Running MESA on a Cluster

Installing and Running MESA on a Cluster

As a real-life example of using MESA on a cluster, here's how one would do it on the Berkeley astro cluster.

1 Download


1.1 Go to your scratch space

cd /clusterfs/henyey/<username>

1.2 Download MESA

svn co -r 4631 mesa

1.3 Download & untar the SDK

wget --user-agent=""
tar xf mesasdk-x86_64-linux-20121106.tar.gz

2 Edit your .bashrc

export SCRATCH="/clusterfs/henyey/jwschwab/"

# for MESA
export MESA_DIR=$SCRATCH/mesa
export MESASDK_ROOT=$SCRATCH/mesasdk
source $MESASDK_ROOT/bin/

3 Compile MESA


3.1 Compile the main program

Make sure to either source your .bashrc or log out and back in. Now running the ./install command should work

3.2 Make your work directory

As usual, make a copy of the work directory. Since you may need a decent amount of space for the logs, this should probably live on your scratch space. Once you have that set up the way you normally would (run ./mk, set up your inlist, etc).

4 Run MESA


4.1 Make a batch file

On one of these clusters, the way you run a job is by submitting one to the batch queue system. There's specific info about this on the Henyey bSpace page. In general, you need a batch file that describes the job you want to run. My MESA ones look like this

#PBS -M <email_address>
#PBS -m abe
#PBS -l nodes=1:ppn=8,walltime=10:00:00
#PBS -q henyey_batch

# this should be equal to the value of ppn
# there are 8 cores on each henyey node


./rn > rn.out

Things that you might want to adjust are the job name (-N) and the job time limit (walltime). The -m & -M flags have the system send you email when your jobs start/stop. And then of course, you may want to do something different with your ./rn command, etc.

4.2 Submit

Supposing you named your batchfile then


Then you can see where are in the queue by typing