I have recently set up a BOINC server for local, processor and data intensive processing. Originally, I thought that I could run the entire project off one server, but now it seems that the work load is too much for just the one server to handle. To increase server capacity, I would like to run the work unit generator, work unit assimilator and data server in parallel on multiple machines. I have searched this site and google for help on doing this, but there just doesn't seem to exist an easy to follow guide (like the project cookbook) on how its done. The MultiHost page says that it is possible, but does not give any complete examples.

Could someone walk me though the steps needed to run the work unit generator, work unit assimilator and data server in parallel on multiple machines. Using a project with the legacy wrapper app as an example would be helpful.

Thanks,
Omar

You'll probably want to share the top level directory of your project between the systems with NFS, so they get the same config.xml and subdirectories.

Then add the name of the secondary host to config.xml for each daemon, and add a crontab entry on the other host to run `start --cron` every 5 minutes. Only the tasks and daemons for that host will be run.

Those are the two main things.

---

-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Great, thanks.