Passive check capabilities

[gergelynemeth]

Hello,

I know that there is a recommendation (per NagiosXI) for metrics, that a new NagiosXI should be considered for about every 20.000 metrics.
Is there such recommendation for a federated monitoring infrastructure? How many passive checks can a NagiosXI handle if it gets all the metrics from NRDP clients?

Theoretically like this:

(VM) (NagiosXI) -> (NagiosXI)
Client -> NRDP Client -> NRDP Server

We're curious about the NRDP Servers capabilities.
Currently we're using a physical server with 40 CPU, 512GB RAM and high-performance SSD (also using ramdisk), which can handle about 50.000 metrics (mostly 5-15 minutes check time, all active checks). Would it increase when every check is received passively through NRDP?

Thank you!

[benjaminsmith]

Hi,

Currently, we're using a physical server with 40 CPU, 512GB RAM and high-performance SSD (also using ramdisk)

That's a pretty good setup. What is the current check load and how is the system performing? Please share the system profile and I can review the logs.

To send us your system profile.
Login to the Nagios XI GUI using a web browser.
Click the "Admin" > "System Profile" Menu
Click the "Download Profile" button

Every one of these large environments is unique so it's very hard to say with definite certainty how well the system will perform at 50K checks. If you have passive checks set up and spread out the check intervals as much as you can, I've seen systems go beyond 50k. It also depends on the type of plugins you are using as some take more resources to execute.

The issue here and why we recommend spitting the environment into multiple servers out at 20k is the time and cost of maintaining a large environment. At some point, you run into diminishing returns on your time.

Some Helpful Documentation:

Distributed Monitoring Solutions

Configuring Passive Services in Nagios XI

Maximizing Performance In Nagios XI

[gergelynemeth]

Currently this is a theoretical question. We're thinking of changing our infrastructure to a federated structure. But it will be only useful for us, if the central NagioXI can handle more passive checks (recevied through NRDP from other NagiosXI servers), than the current.

The average load is about 2-5. And we only see some spikes when a lot of users are polling the Web Interface.
For a rough guess... if there is currently 50k active checks handled, would be about 100k passive checks be handled with the same performance?

Thank you in advance!

[benjaminsmith]

Hi @gergelynemeth,

I wish I had a definite answer here but there are a number of variables in play here. For the Nagios process itself, that might be a plausible assumption. However, there is still the processing of performance data if that's being returned with the check. Additionally, some checks like SNMP take more resources to process internally.