I'm trying to work out a cluster for personal experimenting. I think I have a pretty good understanding on how to set it up (router servers, nodes, and IP configs), but how would I manage the actual data on the nodes?

So if it's a cluster balancing http service, if I make a change to the site, how can I make the change once and have it propagate to all nodes? I feel like I'm missing some critical element of a load balancing cluster.

Thanks!

We do the following, we have one web server designated as the "upload" server. Any changes to the website are uploaded to a folder structure on that machine, let's call it /home/upload/html

We then have a sync script we run manually on each of the three main web servers in a specific order, the first sync script is on the "local" machine with the update, this then does a local rsync from /home/upload/html to /var/www/html, then other two machines then do an rsync from the first's /var/www/html

You could always fully automate this process and have the other servers rsync on a timed basis (hourly, 5 minutes, 1 minute, your choice) from the "upload" server.

Indeed, if you were to automate it you could run a single "PUSH" rsync from the upload server to the other servers and have this on a cron job or trigger it manually from a shell.