[Nagiosplug-help] Interesting problem when checking remote host

[Andreas Ericsson]

Andre Plante wrote:
> I am trying to setup some monitoring on a remote host.  On the remote
> host, I have installed the nagios-plugins, version 1.4, as well as nrpe,
> version 2.0.
>  
> When on this host, I can run all my checks locally and they run fine.
> The results that are returned when I run the commands in the nrpe.cfg
> are accurate.
>  
> However, when I try and run a check_nrpe command from the monitoring
> host, I get incorrect values back.
>  
> For example:
> On the monitored host, I do:
> /usr/local/nagios/libexec/check_procs -w 1:3 -c 1:3 -C crond
> and get:
> PROCS OK: 1 process with command name 'crond'
>  

You're most likely running this as root.

> On the monitoring host I try:
> ./check_nrpe -H 216.x.x.x -p 5666 -c check_crond_proc
> and get:
> PROCS CRITICAL: 0 processes with command name 'crond'
>  

NRPE is likely NOT running as root.

This is caused by a kernel-patch which hides part of /proc from non-root 
processes. It's default in Openwall GNU/Linux, although some other OS's 
(notable those with SELinux) have something similar in place.

> I have tried this with and without the port number (to ensure that it is
> not the culprit, and get the same result back.  I know that the
> communication is working because when I try it with total procs, I get
> in the 40 range on the monitored host, but only 4 when on the monitoring
> host.
> Any help is greatly appreciated!
> Thanks,
> Andre
>  
> Andre Plante
> Network Administrator
> DINMAR
> 400-495 March Rd.
> Kanata, ON  K2K 3G1
> email: aplante at dinmar.com
> Phone: 613-287-3130 ext. 477
> www.dinmar.com
>  
>  
> 
> 
> 

-- 
Andreas Ericsson                   andreas.ericsson at op5.se
OP5 AB                             www.op5.se
Lead Developer