Monitoring win server memory "real time"

[roadrunner]

Hi,

I'd like to monitor and keep history of several win server. So far, i have my nagios running, and monitoring one single win server. But all I have, so far, is a screen that says how many alerts there was. I actually want to know on a 24H interval what is the memory usage almost in real time... i don't need a real 'real-time' graph, but i'd like to have a 1 or 2 minutes interval consumtion graph... how can i manage to get this ?

Thanks
R

[jsmurphy]

This tutorial should help you configure graphing: http://xavier.dusart.free.fr/nagios/en/nagiosgraph.html if you are new to Nagios that guy has some fantastic tutorials that are worth checking out.

I probably wouldn't configure 1 - 2 minutes intervals, depending on how much you are monitoring that could put some serious load on the Nagios server... not to mention most CPU and Memory checks return the average of the last 5 minutes of utilization. So maybe configure the CPU and memory checks to poll at 5 minute intervals.

[roadrunner]

Hi,
it seems to fit perfectly my needs... thanks a lot for the link... i appreciate !
R

[roadrunner]

Hi again... Things are not working that fine with my nagiosgraph... My general conf seems to be ok, but i have a problem when i look at my nagiosgraph.log... i keep having these messages :
Mon Oct 15 13:13:46 2012 p1.pl error RRDs::update ERR /var/spool/nagiosgraph/rrd/localhost/HTTP___time.rrd_min: conversion of '0,000000' to float not complete: tail ',000000'
Mon Oct 15 13:13:46 2012 p1.pl error ds = [
'/var/spool/nagiosgraph/rrd/localhost/HTTP___time.rrd_min',
'1350299614:0:0,000000'
];

I've seen a message that says i was to add LC_ALL=C LANG=C before the check_nt command... so i did modify the command.cfg and added it to the line and so have something like this :
command_line LC_ALL=C LANG=C /usr/lib/nagios/plugins/check_nt -H $HOSTADDRESS$ -p 12489 -v $ARG1$ $ARG2$

I don't have error when checking the nagios.cnf, but i keep having the error in the nagiosgraph...

Does anyone have any leads to solve this ?

TIA
R

[roadrunner]

Ok, i give an answer to my problem... actually 2 answers... First of all, i figure out why and how my correction with LC_ALL=C LANG=C didn't work... i actually had twice the same server monitored, with two different names ! yes I know... my mistake ! Still my rrd folders were desperatly empty... and there was another rrd folder that was filled with datas... fedora yum install did change a lot of directory, so i certainly was wrong when giving the rrd directory to the config file... now i have my memory graph that comes to live...

But I still have erratic behaviours with the collection of data... some times i have graphic results, some times not. I had a 20 minute graphic line without problem, then a smal gap, then another graphic, and another gap... to be clearer, i have a discontinued line when watching my memory usage graphic... And I stille have the errors LC_ALL was supposed to correct...

Regards
R

[jsmurphy]

To be honest we are beginning to creep outside the scope of my knowledge, I've never really encountered sufficient problems with graphing to have a particularly good understanding of how all the little parts works or how to debug it.

You could post here: http://sourceforge.net/projects/nagiosg ... rum/394748 you are more likely to find help.

Worst case scenario you could also look at pnp4nagios which is another popular graphing solution, though it is a little more complex to configure: http://docs.pnp4nagios.org/pnp-0.6/start

[GaWd]

I've never had much luck with nagiosgraph, honestly. n2rrd is also extremely complicated to deploy.

I have had excellent luck with pnp4nagios, however. It deploys rather quickly, and produces graphs immediately.

[agriffin]

I haven't encountered this problem before, but I do know that there's no point in setting LANG when you've already set LC_ALL. On the other hand, setting LANGUAGE instead of LANG might actually do something.

[roadrunner]

Ok,

I just saw that i've configured 2 computer with the same IP.... my mistake... instead of monitoring 2 different computers, i was monitoring one with two different names !!! thanks to memory alerts that are synchronous, i saw that... don't know if it's the reason for my errors, but it could be a lead.

I'll see this in a few minutes...