[ILUG] tip of the day

[Colm MacCárthaigh]

On Fri, Dec 20, 2002 at 04:54:36PM +0000, John P. Looney wrote:
>  If you are looking after machines, check out nagios. It's quite good - it
> does all sorts of monitoring, so you don't have to. You can set it up to
> monitor all your servers, and email you when things go bad - like the
> exchange server stops responding to IMAP queries..
> 
>  More importantly, most of the plugins have "warning thresholds"; for
> instance, I monitor all my machines to make sure that they have at least
> 20% free on all disk partitions. It's nice to get an email saying the
> webserver is 85% full, because someone left a PHP page echoing a few dozen
> kilobytes to disk every time it's hit....before the website falls over.

How are you checking the disk space on remote machines ? The only
nice way I could think of doing it was creating an unencrypted ssh key 
which limited what could be run to one script, e.g:

no-port-forwarding,no-X11-forwarding,no-agent-forwarding,no-pty,command="~/bin/stats"
<rest of ssh public key>

and giving a netsaint plugin access to the private key. Havnt gotten
around to doing it, would be intresting if there's a better, secure way ? 

>  However, it uses snmp to do much of it's work, and on my little raq3 that
> does this monitoring, the load is quite high, as snmpget is a very
> expensive program to start off (something like .45 of a second), as it has
> to read through around a meg of structured text MIB files, every time it
> starts.
> 
>  However, Mr Brady pointed me in the direction of;
>     http://freshmeat.net/releases/107146/
> 
>  which is an snmp proxy for nagios. I changed my polling time from eight
> minutes to three, and the load went from three to 0.0 - for the first time
> in months!

Never noticed load, though ours is on a P4 2.4Ghz Xeon with a Gig of
Ram :)

-- 
colmmacc at redbrick.dcu.ie        PubKey: colmmacc+pgp at redbrick.dcu.ie  
Web:                                 http://devnull.redbrick.dcu.ie/