[olug] Open source monitoring system?

[Christopher Cashell]

On Wed, Apr 15, 2009 at 4:33 PM, Jay Hannah <jay at jays.net> wrote:
> If my Google-fu is up to snuff this thread hasn't happened on this
> mailing list this year or last.
>
> Our Big Brother 1.9e from 2004 is quite long in the tooth.
>
> If you were going to start over today what open source monitoring
> system would you install to watch your servers / disk / processes /
> etc and do paging / email notifications, and escalation?

I would actually take this in two parts, and depending on how many
machines you're dealing with.

Big Brother, especially an older release, sucks.  There are a couple
of projects that began as Big Brother addons that have since outgrown
it to become independent replacements (Hobbit and Big Sister being the
best known).  The great thing about these projects, is that they're
protocol compatible with Big Brother.  And, as much as I like Nagios,
it can be a beast to get setup initially.

So, assuming you have a reasonable number of machines (and not too
many custom serverside scripts), I'd start by doing a replacement of
Big Brother on the server side with Hobbit.  I wouldn't touch the
clients or client machine configs.  I'd just drop Hobbit in and let it
continue monitoring everything that Big Brother is, only doing a
better job of it.

Then, after Hobbit was monitoring the existing setup, then I'd start
working to get a Nagios replacement up and running.

The primary case where I'd consider something other than Nagios is if
the majority of the devices you're monitoring are network devices or
appliances, and your primary means of monitoring and communication
would be SNMP.  In that case, I'd be looking more towards a
Zabbix/Zenoss/OpenNMS/Cacti/etc solution.  Nagios can do SNMP
monitoring, but that isn't its strength.  Some of the others are more
oriented around SNMP and do a better job with it.

> j

-- 
Christopher