Nagios Support Forum

Hello, Is there a document showing the traps per second supported and/or tested in lab environments? I have a difficulty preparing the sizing of an installation and cannot see if more than 50 traps/second will be achievable. (steady flow - every second of every minute of every hour of every day - yes it is despairing). Sorry for asking here - I checked in a lot of places and turned empty.

3 sizing questions in 2 days... we really need to make a sticky that says "we don't know"

I know this doesn't read like a sizing question, but the reality is that is what it is. What you're asking us is how big (if at all) a system needs to be to handle 50 perl plugins a second. To nagios that's what the workload would look like. That's a BIG workload. Plan on lots of CPUs, memory and lots of IO to your database. You're talking a scale also that I've never tried to size for so I'd suggest modeling it in a VM and slowly ramp up to that level and see how the system reacts.

Ignoring those basics - another big concern would be that almost always a trap is indicative of a problem, meaning a notification. A service check that results in a notification and/or handler in Nagios can consume limitless amounts more resources than a check that is not acted upon. This is due to notification actions spinning up for some random number of users (memory heavy) and possibly event handlers also, on-demand checks etc.

I'm not saying that your use case is necessarily mimicking that, but it is something to be wary of.

The short answer to your question is that I've never seen a functioning environment handling that many perl service checks (3000/min) so I don't know how nagios will react. Maybe someone else will have more experience.

Is your question about your Nagios server RECEIVING or SENDING 50 traps/second ?