Monitoring OSPF

[Ruben Laban]

On Friday 09 September 2011 at 12:52 (CET), Ondrej Zajicek wrote:
> On Thu, Sep 08, 2011 at 09:52:02AM +0200, Ruben Laban wrote:
> > Hi list,
> > 
> > I was wondering if there's a sane method to keep a realtime eye on OSPF.
> > More specifically, getting notifications for link/state/topology/etc
> > changes. So far all I could think of is using birdc(6) every x
> > seconds/minutes and comparing the output with the previous run to detect
> > changes.
> 
> There is no dedicated notification channel. The simplest thing is to get
> output of 'show ospf state' every x seconds. Or if you want, you could
> enable debug events logging and use routing the table recalculation
> message as a trigger for 'show ospf state' (but that seems unnecessary for
> me).

That's basically what I had thought of myself as well. Glad I didn't overlook 
any other obvious methods ;-)

> For monitoring, i have a script that logs the state, sorts that
> appropriately to hide irrelevant changes, compares with the previous one
> and possibly sent the diff (attached, just a quick hack).

This script could use some polishing indeed. But it's a nice start.

> I have also some scripts that took 'show ospf state' output and expected
> network description and converts that to the SVG image with reachable,
> broken and unreachable links differentiated by color, but it is not
> really finished.

I'm currently working on something similar: a perl script that also parses 
'show ospf state' and uses graphviz/neato to turn it into a nice diagram. 
Annotating different kinds of links (vlan, tunnel, healty, broken, etc) is 
still on my todo list as well. What do you use to generate the SVG? As of all 
layout engines I've tried, none really manages to create clean graphs for our 
OSPF topology (currently about 15-20 nodes).

Regards,
Ruben Laban