Monitoring with Ganglia

Chapter 5. Managing and Extending Metrics

Brad Nicholes

Daniel Pocock

Jeff Buchbinder

In this chapter, we describe the various ways in which the Ganglia monitoring environment can be extended. Primarily, we discuss two ways in which to extend Ganglia, including the development and deployment of additional metric modules, and the use of a standalone utility called gmetric. Either is suitable, and you should use whichever approach works best for your environment. If you are someone who likes to get down and dirty in source code, developing a gmond module might be the way to go. On the other hand, if you just need a quick way to introduce an additional metric in your environment, gmetric is the perfect utility.

gmond: Metric Gathering Agent

The Ganglia monitoring daemon (gmond) is a lightweight agent whose primary purpose is to gather and report metric values. The first step in monitoring any system through Ganglia is to install the gmond daemon on each machine. Once installed and running, this daemon uses a simple listen/announce protocol via eXternal Data Representation (XDR) to collect and share monitoring state information with other gmond services within a cluster. In the Ganglia monitoring world, a cluster is defined as a group of gmond services that are all listening and sharing data with each other. A Ganglia cluster might consist of anything from a single gmond node to many nodes, all talking to one another. In the default configuration, each node listens and talks on a single ...