System monitoring

[cdevers at pobox.com (Chris Devers)]

On Thu, 22 Apr 2004, Cole Tuininga wrote:

> On Thu, 2004-04-22 at 15:21, dsr at tao.merseine.nu wrote:
> > Have you evaluated mon?
>
> Nope - do you use it?  If so, what do you like about it?

My company uses Mon, and people seem to like it.

The thing to keep in mind is that Mon, like some of the other network
monitoring packages you'll find, sits on top of the SNMP protocol. SNMP
is nice because it lets you query all kinds of statistics about any
given network attached host -- where "host" in this context includes
things like printers, routers, etc along with more conventional things
like servers, desktops, and laptops.

So what you really need may be a toolkit that sits on top of SNMP and
provides you with various reports about what's going on with your
network. Anything that tries to do this without using SNMP is probably
going about the problem the wrong way.

Here, we have Mon filling this role by emailing is with things like
status reports, warning & error reports, pages with critical errors, and
so on. Additionally, it provides a web interface that informs us what's
going on -- what systems are acting up, what recent activity we've been
having on different hosts, etc.

Mon was set up here before I started, so I can't really address the
issues that may have come up in building the infrastructure for it. My
main complaint is that the number of email alerts can be overwhelming,
and picking out the relevant stuff can be a lot of work. I don't know if
this is a general Mon issue or if it's a side effect of local policies.

Learn more about Mon here:

    <http://www.kernel.org/software/mon/>


Another tool worth looking at is MRTG, which provides a web interface
with very nice activity graphs for all your network devices.

    <http://people.ee.ethz.ch/~oetiker/webtools/mrtg/>

We use MRTG here as a complement to what Mon provides.



-- 
Chris Devers