Is Fusion the answer for us?

[jbennett]

We currently have a single Nagios install monitoring a very large number of hosts and services. I am utilizing NRDS in order to help take the load off of the server, as well as ramdisk, but I am still running into issues where I must stagger the checks being sent to the server over an unreasonable time frame in order to prevent the server from locking up.

We currently monitor 941 hosts and 22584 services.

A very large number of these services are checked via NRDS. I have to stagger these checks from 3 to 10 minutes in order to spread the load. However, this is really not desired due to the nature of our systems.

I should also note that I would ideally like to add at least another 1000 hosts to this configuration in the form of a simple ping. I would also like to be able to add more service checks on existing hosts, but am limited currently as we cannot process everything in a timely manner on our current server.

Would splitting this set-up over multiple boxes and utilizing Fusion be my best bet?

[abrist]

Yes. Otherwise you would need to investigate offloading as many services as possible: ndo, mrtg, mysql, postgresql.

[jbennett]

ok - we might consider offloading mysql anyways.

However, I wanted to know if there was an easy method to determine processing and memory needs? I am not a server admin by trade (I just pretend) and I will need to provide VM requirements to the server team in our company.

[tmcdonald]

Well, that all depends on what type of checks you are running. Running 20,000 check_dummy services is a lot easier than 20,000 check_esx3 services. What would you say are your most common checks?

[jbennett]

NRDS checks on the hosts themselves and then sent to the Nagios server. These checks are custom. They are checking a .txt file on the remote servers for specifics.

[lmiltchev]

However, I wanted to know if there was an easy method to determine processing and memory needs?

You can use the following document as a guide:

http://assets.nagios.com/downloads/nagi ... ements.pdf

NRDS checks are passive. They won't put lots of stress on system resourses.