[Wolves] Icinga (Nagios)

[Andy Jewell]

From: wolves-bounces at mailman.lug.org.uk [wolves-bounces at mailman.lug.org.uk] On Behalf Of Adam Sweet [adam at adamsweet.org]
Sent: 18 May 2010 14:11
To: Wolverhampton Linux User Group
Subject: Re: [Wolves] Icinga (Nagios)

Peter Cannon wrote:

> OK in my usual 'lazy git Pete' manner I wanted, for example, to be able
> to initially say to someone "Download this from
> www.honest-turpin-support.com/nrpe.zip(whatever) install that".
>
> Write the configuration template in Nagios at my end, start getting info
> from the remote machine. Sort of a very very low level support type
> thing "OMG Chris_P's server has just fell over" Ring ring, ring, ring,
> "Hello Chris? Your Servers fell over mate" "Ooh has it? Oh yeah I cant
> get on xxx drive I love the way you know before we do Pete" Kissy kissy
> hug hug.
>
> I'm not talking a simple ping service here it would be nice to be able
> to say "Chris_P your FFF drive is getting full" or some of the other
> info I know you can get.
>
> As for fixing the fallen server that's another matter, the point being
> (And I know this is not going to happen) Ideally I wanted to be able to
> have a daemon or service running on the remote machine stick the IP in
> Nagios, management machine without and ssh or firewall crap to deal with.

All of that is perfectly possible but you can't monitor anything if the
remote firewall won't let you in. They just need to know the IP you're
coming from, the port you're connecting to (where applicable) and the
protocol you're using - TCP, UDP or ICMP. If you don't know your
protocols and port numbers, now is a good time to read up. Look at
/etc/services for a run down on which services use which ports and
protocols.

If you don't want to do checks by SSH you can do checks by NRPE as has
been discussed. I forget whether I saw anybody tell you to put
NSClient++ on any Windows servers you want to monitor and check them
using NRPE.

For publicly exposed services like HTTP, SMTP, IMAP, POP3 and pings you
don't need to do any SSH or NRPE stuff, you can already see that stuff
from the Internet at large. It's when you need to check variables local
to the remote machine you need to start doing that kind of thing because
you need some way of asking for the amount of free disk space or the
load average. The alternative to checking by SSH or NRPE is to get jiggy
with SNMP, but that's another story if you're still grappling with
Nagios itself.

Nagios takes some getting used to when you're first starting up, think
it took me 2 weeks the first time, but once you're up, you can pick up
speed quite quickly.

Use the following command to check your config before reloading Nagios
after making changes so you get an idea if there are any problems in

nagios -v /etc/nagios/nagios.cfg

Of course, specify your nagios binary and your master config file it
those in the example don't match those on your test machine.

Regards,

Adam Sweet

--

http://blog.adamsweet.org/


_______________________________________________
Wolves LUG mailing list
Homepage: http://www.wolveslug.org.uk/
Mailing list: Wolves at mailman.lug.org.uk
Mailing list home: https://mailman.lug.org.uk/mailman/listinfo/wolves

I also forgot to mention (and Adam's posting reminded me) that if you want to check Windoze *logs* the check_mk agent can do that. I wasn't really thinking in terms of Windoze boxes. 

The check_mk agent also has other advantages in that it reduces check traffic by returning all check results at once in response to one check request (it then submits these back to nagios via "passive" checks - see the manual).

Andy D'Arcy Jewell
SysMicro Linux Support

T:  +44 (0) 844 991 8804
M: +44 (0) 7961 605631
F:  +44 (0) 844 357 7020
E:  andy.jewell at sysmicro.co.uk
W: www.sysmicro.co.uk

CRN THE CHANNEL AWARDS 2009 WINNER
SysMicro named CRN’s Editor’s Choice for Emerging Business of the Year 2009, recognising SysMicro for our considerable growth, specifically in Enterprise Solutions.
________________________________________