Re: [ns] infinite loop problem in AODV, both in 2.1b7a and snapshot?

[Felix Ko <felix.ko@mchp.siemens.de>]

Hi all!

As far as I know, AODV doesn't delete routes at all.
If they are expired, they are just _downed_.

Another thing: In a normal simulation, a node isn't resetted at all
or am I wrong?! (At least *I* am not resetting them during the sim!)

Bye
Felix

Roland Klabunde wrote:

> > Can anyone please explain exactly what condition causes the AODV to
> > run into the infinite loop? I remember a posting to this list some time
> ago
>
> I remember the AODV loop trouble as _not_ related to NS. I've heard, that
> guys of a scandinavian university found AODV's potential loop, if one
> station, which had a complete route before, is resetted in between and
> restarts from "zero knowledge". I remember furthermore, that this has been
> taken into account in the latest RFC (draft-ietf-manet-aodv-08.txt, March 02
> 2001), implementing a DELETE_PERIOD timer. Each node on reboot waits now for
> this period and does not respond to route packets. After this period none of
> its neighbors will treat the rebooted node as active next hop.
>
> I'm currently not sure, whether this (new) behavior is implemented in
> 2.1b7a.
>
> Regards
> Roland