I've been playing with Nagios -- I've been meaning to do so for a very long time -- and have had some fun with it.

But can anyone here suggest plugins/additions/alternatives that would help me achieve what I want to do?

Basically, as well as being emailed when something unusual happens, like the qmail queue increasing more than usual, I'd like to see charts of the qmail queue size over 24 hours/7 days/1 month/1 year.

I'd also love to see a chart of spamassasin/qmail-scanner stats showing number of emails received and how many were dropped and how many tagged, again over 24 hours/7 days/1 month/1 year.

I can only seem to find two qmail plugins for Nagios and both seem very simple. The only spamassassin plugins I found were related to checking that spamd was up rather than giving me stats.

I know Nagios is mostly to do with monitoring, but for some plugins, for example one related to checking the health of Dell servers and when used in conjunction with an additional pluging - PNP4Nagios - some lovely charts can be generated.


Can someone point me in the right direction please so I can have some fun over the Festive break?

Thanks,

Faris.

Love the way you think! Festive breaks are for server related fun!

BTW, I am also really interested in Nagios. Been on my todo list for quite a while now...

Hehehehehe.

Nagios turns out to be both easy and difficult.

There are RPMs in the rpmforge repo. You just need nagios, nagios-plugins and nagios-plugins-nrpe, the latter being required in order to communicate relatively securely with the nrpe daemon that you'll need to run on the remote machines you want to monitor.

On those remote machines you need to install the plugins and nrpe. HOWEVER, you can't really use the RPMs for those because they are set to require nagios itself as a dependancy. Therefore grabbing the source and compiling them is the only option (not that it is difficult!).

In fact it is probably more sensible to compile everything from source. I was just being lazy.

Some useful links:

http://wiki.centos.org/HowTos/Nagios
http://www.thegeekstuff.com/2008/06/how ... nagios-30/
http://techchorus.net/how-install-and-c ... ios-centos

And some Android apps that can tap into Nagios: http://www.androidpolice.com/2010/06/06 ... onitoring/
CRUCIAL: When entering the URL in the Android app, use https://domain.tld/nagios/cgi-bin and NOT /cgi-bin/nagios and NOT just /nagios

Faris.

I like Zenoss Core for monitoring and graphing, which supports Nagios plugins for monitoring (a lot of ZenPacks are actually based on Nagios scripts). I never understood the fun of plain Nagios (ugly GUI, complex configuration, just monitoring, no graphs).

You mentioned this recently but I think I ended up in the wrong place that time -- All I saw were commercial options costing many $1000s. This looks interesting. I'll investigate. Thanks again Breun.

Faris.

Zenoss Core is open source and free to download, Zenoss Enterprise has some more features and is not that free. They recently moved the open source community to zenoss.org, focusing more on their commercial offering on zenoss.com.

Well, I have Zenoss installed and running, and monitoring three servers, but.....it is driving me up the wall. It is basically the config that's annoying me. snmp should "just work" but I'm not getting any details shown even though the connection is there and an snmpwalk works. I really don't like snmp AT ALL and never did. This is probably to do with the fact that I'm trying to monitor Virtuozzo Containers rather than physical servers.

And OHM Zenoss is sooooo slow. I'm running it on a very old machine with only 512Mb of RAM and a CPU from the days before multicore, but we're talking 10 minutes for Zenoss to even start.

I'm not saying it isn't a brilliant product though. It is light years better in terms of UI than Nagios. But I fear it just isn't for me. I'll keep trying though.

Faris.

SNMP with Zenoss 'just works' for us. Yeah, 512 MB RAM for Zenoss isn't going to be a lot of fun. You'll want 1-2 GB for 2.x and probablty more with 3.x.

Just to update this thread.....

I've had great fun with Nagios and pnp4Nagios (for the charts) and it is currently monitoring all sorts of stuff very nicely.

There are some huge "gotchas" though. Here are some tips:

1) Do NOT install nagios from RPM as it installs things in places that are different to installing from source. This can be a problem when later adding things such as pnp4nagios and other add-ons. Seriously folks, "./configure; make; make install" is not that much harder than "yum install nagios".

2) If you use the check_qmailq plugin please talk to me - I've created a modified version that returns performance statistics (e.g. messages in queue, messages in queue not pre-processed) which you can then view on a chart. Once I figure out who I have to ask permission from/what I have to do I'll add this modified version to the Nagios Exchange.

3) If you use check_qmailq AND you opt to add user nagios to the qmail group in order to get around the permissions problem AND you run the plugin via NRPE on the remote system then you need to add groups=yes in /etc/xinet.d/nrpe just under where it says group=nagios as otherwise xinetd will ignore supplementary groups and it just won't work.

4) My personal preference only: Some plugins are written in perl and need certain CPAN packages. Instead of getting these from CPAN you can usually find an RPM version of the same package in RPMforge.


It is a great shame I didn't get on with Zenoss (mostly due to insufficient hardware to run it on, though I have to say that the version 3 UI is a bit hard to get along with) because I can see the serious potential it offers and how it is more flexible and more sphisticated than Nagios.

I tried getting qmail, spam assassin, and plesk in general working with Cacti as I have a very large cacti install monitoring the server and many other devices already. I had no luck finding a template and am not versed enough to create my own templates or graph templates.

I had alway hoped that someone would create something that would allow me to monitor the email queues, spam assassin stats, http stats, per user bandwidth etc... but so far no one has gotten anything working for Cacti.

Following this thread intently. Tried setting up cacti and failed so far

setting up cacti is easy its getting it to monitor the stuff you want thats not so easy. If you want a quick and dirty way to get cacti working and play with it? check this out. http://cactiez.cactiusers.org/

@faris: You got a modified check_qmailq with perf stats? Does it also return messages sent / minute? Searching such a plugin ... but no luck yet
Maybe you can contact me ...

bye from Austria
Andreas

No, just number of messages in queue/not pre-processed (or something -- useful to alrt you to a spam outbreak only)

Hey faris, here's an idea:
https://www.monitoringexchange.org/inve ... _munin_rrd
I'm actually using a munin plugin in Nagios to remotely monitor various server resources such as load average, mail queues, generic network connections usage, mysql status and so on. You'll have to install munin-node on the monitored server, configure the needed plugins and secure it against unauthorized access.