Nagios Server Baseline

[joe1871]

Hi,

I have a fairly large Nagios implementation that is currently monitoring > 500 hots with >800 Active checks. I have started to see RSZDT pprocess warnings on the server, and I also am just curious about what a typical baseline number of hosts/checks would be for a single server. Is there any doc or can anyone tell me what they think the baseline might be for a Nagios server? I realize it will depend signififcantly on the server itself, CPU, # cores, RAM etc. Assuming all of that is at a comfortable level, is there a target number of hosts per Nagios server? Do I need to break up my installation into more than one server for capacity issues, and if so, is there a standard for Nagios pooling?

Thanks in advance!

Joe

[mguthrie]

I wouldn't say there is a "typical." I'd say on average you wouldn't want to run more than 10-20 thousand active checks (hosts and service checks combined) in a single Nagios core instance. Take a look at your Performance Info page and watch the check latencies. If your average latency is over 10 seconds then you're pushing your server too hard with active checks. Many people branch into distributed monitoring between 10 and 20 thousand service checks, depending on the power of your machine.

This can vary greatly on what checks are being used, what addons you're using, if you're graphing performance data, etc. Check out the Nagios Core manual on sourceforge and look up the section on performance tuning. Let me know if you need the link.