Hi!
We are planning to start monitoring our servers by nagios (or nagiosxi), but we have reasons why we want to do it by check_by_ssh.
There is written in document: "This method will not be covered here, but can result in high load on your monitoring server if you are monitoring hundreds or thousands of services. The overhead of setting up/destroying SSH connections is the cause of this. "
How big environments have you been able to monitor with one Nagios instance using check_by_ssh? I know that of course this depends on amount of servers, services and polling intervals, but some kind of vision would be nice.
I have also read, that using Controlmaster -option in SSH would increase the performance.
The environment that we will monitor has about 500-1000 *nix servers and every server will have about 5-15 services to monitor.
Thanks in advance!
check_by_ssh performance
Re: check_by_ssh performance
"It depends"
What are the specs of your box? Do you plan to use any Nagios additions that will cause additional overhead (i.e NDOUtils and Graphing or mapping)? How many users actually using the Nagios GUI?
I think with that information we can sort of have a guess at it... but Nagios is a very modular application and it's hard to say with any degree of certainty unless it becomes obvious your box is confidently under or over specced for the job at hand.
What are the specs of your box? Do you plan to use any Nagios additions that will cause additional overhead (i.e NDOUtils and Graphing or mapping)? How many users actually using the Nagios GUI?
I think with that information we can sort of have a guess at it... but Nagios is a very modular application and it's hard to say with any degree of certainty unless it becomes obvious your box is confidently under or over specced for the job at hand.