From Debian Clusters

About Ganglia

Ganglia is a monitoring system out of Berkeley designed specifically for clusters and grids. It can be used to monitor workload at a glance. Unlike Nagios, which runs by default on a single host machine, roles are split amongst machines with Ganglia. The head node or web server can run the Ganglia Meta Daemon (called "gmetad"), and each worker node runs a Ganglia Monitoring Daemon ("gmond").

If your web server runs on your head node, this is easy. Otherwise, you may want to install it on the web server instead of the head node.

The tutorial for Ganglia is broken into three parts:

• Installing Ganglia
• Configuring Ganglia: the host node (gmetad)
• Configuring Ganglia: the client nodes (gmond)

References

• Ganglia Monitoring System
• Ganglia Wiki: Readme