Going data intensive?

As a result of a few emails and some long running thoughts, I have finally set up a small, heterogeneous cluster of Raspberry Pis. Using the 2s and 3s that I have from other projects, I put the machines together with a small switch and installed MPI with the Python bindings to test some scripts.

I largely followed PJ Evans’s MagPi article, Build a Raspberry Pi cluster computer, and James Mackenzie’s Headless Raspberry Pi Setup to add in ssh access without logging into the machine. The instances all have Raspbian Lite installed as I need a minimal install. I did have to make a small changes to Evans’s excellent article though.

I found that the code:

mpiexec -n 4 –hosts 10.0.0.1,10.0.0.2,10.0.0.2,10.0.0.4 hostname

needed to be:

mpiexec -n 4 –host 10.0.0.1,10.0.0.2,10.0.0.2,10.0.0.4 hostname

So why build one? One of the areas that I would like to do some further research into is Data Intensive Humanities but other things need to happen first.

No Comments

Leave a Reply

Your email is never shared.Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.