[olsr-dev] Help diagnosing problem

[Ignacio García Pérez]

Hi,

I have the following setup: 20-some nodes connected by point-to-point SLIP
lines. As you see, the nodes don't move and the links never change (unless a
wire is broken), so OLSRD is mostly used to take care of automatic network
configuration.

There is one "central" node. I quote "central" because of course to OLSRD it
is just one more node. But I can reach this node remotely through broadband,
while I reach all others through the SLIP lines. This means that if I touch
anything in any node but the central one, and it does not work, I'll lost
any possibility of remotely fixing it. So I must be carefult and I'm limited
to experiment only at the central node.

All nodes were running a slightly modified 0.4.5 OLSRD daemon. The
modification were some bugs I fixed and some changes to force olsrd to not
ignore point-to-point interfaces and interfaces without a broadcast address.

Since the SLIP lines are running at 9600 bps, I have increased the packet
transmission period and the timeouts to avoid sucking too much of the
available bandwidth. HELLOs are sent every 10 seconds, and the HELLO
expiration is set to 30 seconds.

Recently the central node (which is a normal PC instead of an embedded
system) crashed, so I had to replace and reinstall all the software. I
installed OLSRD version 0.4.8. I could not use the old configuration file
since format has changed. Just created a new one and set up the timeouts as
required.

Eveything seemed to work fine at first, but now that it's been running for a
couple o weeks, I've detected some problems. The OLSRD daemon in the central
node seems to be dropping the routes to some nodes once in a while (i.e.
once each two days). It took me quite long to realize that it was the OLSRD
daemon due to some bizarre side effects of my routing setup.

The funny thing is that the dropped nodes are usually in the only way to
reach other nodes, and those nodes are not dropped:

C-----1-----2-----3-----4-----5

In this "linear" setup, for example, route to node 3 is dropped, but routes
to nodes 4 and 5 remain and those nodes can be reached normally.

Anyway, this is my question: does anyone know of some bug in 0.4.5, in
0.4.8, or some version incompatibility that may be causing this?. I just
launched the OLSRD daemon in the central node in debug mode, so I should
have some logs to check out in a week, but meanwhile any directions will be
greatly appreciated.

Regards. Nacho.



-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.300 / Virus Database: 265.6.12 - Release Date: 14/01/2005