[Olsr-users] Strange OLSR behaviour

[ZioPRoTo (Saverio Proto)]

> The
> problem is that those two nodes ("glinska-9" and "fri") are cut off as
> the link from "stefanova-15-ii" is flapping.
[..CUT..]
>> And tap0 interface has LinkQualityMult default 0.80 added. And yes, we
>> have the same IP addresses on all interfaces (so on "gerbiceva-49"
>> both tap0 and wl0 interface has the same address).

I'm replying to this email on the train, so I have Internet connection
to go on the web site and get the details of the node .... but my
quick feeling is:

It is all on the same radio channel ?
Are you sure it is a OLSR problem ?

Even if the link between "glinska" and "fri" was good before, when you
added "stefanova" node with a directional antenna that could add
interference to the preexisting links.

Could be that "fri" hears very strong signal from "stefanova" that is
directional and this disturbs the other preexisting links. Moreover is
difficult to estimate this behaviour because probably the interference
happens only when "stefanova" transmit actual traffic ...

what you think about ?

anyway if you want to exclude the radio part it is quite easy to
reproduce this scenario in Netkit[0]. It is just 4 nodes. You can run
4 UML OpenWRT virtual machines with the same version of OLSR, attach
them in the topology you specified, and emulate packet loss using tc.
This way you can really debug if the problem is OLSR and not the
random radio interference.

Saverio

[0] - wiki.netkit.org