How does Nagios monitor the health of its agents?

[gormank]

Forgive me if this is an ignorant question. I'm new to Nagios.

In other monitoring systems I've worked with, there were heartbeat alerts sent by NEs, or by agents to let the system know that things were still working.
Ping tells us the server is on the network, but if an agent dies or locks up, how does Nagios know?

Thanks

[tmcdonald]

If you are using an agent to monitor a machine and that agent dies, then all checks run through that agent will go into a CRITICAL state.