[OLSR-users] unstable routing tables with 0.4.8

[Andreas Tønnesen]

First of all, as Christian points out in another mail, do _not_
run with a debuglevel of 9! All that I/O causes a lot of CPU usage.
Second, make sure your plugin is not CPU consuming. Remeber
that it runs in the same thread as olsrd itself. If a plugin uses
any kind of blocking or timeconsuming I/O without doing that in a
separate thread, there might be serious trouble.

Regarding the bugfixes, one perticular fix comes to mind. There
was a problem with the order of route addittions so that routes
in some causes, was added using a nexthop that was not yet added
to the routing table. This would cause the kernel not to add the
route.
However, I doubt this is what causes your problems, trying 0.4.9
was merely a suggestion to rule out this, and other, fixes. If
this bug was the problem, then olsrd would display all nodes, but
the kernel table woul not show them.

Try with debuglevel 0(or 1).


- Andreas

Franca Delmastro wrote:
> Thank you for your suggestion, but I have another problem:
> I developed a plugin for the version 0.4.8 that is strictly necessary for my
> application.
> I wouldn't spend a lot of time porting the plugin on the last version  of
> olsr.
> 
> Could you tell me please what kind of bug have you fixed related to this
> scenario??
> 
> 
> Thank you
> Franca
> ----- Original Message ----- 
> From: "Andreas Tønnesen" <(spam-protected)>
> To: "OLSR discussion and development" <(spam-protected)>
> Sent: Wednesday, August 03, 2005 12:33 PM
> Subject: Re: [OLSR-users] unstable routing tables with 0.4.8
> 
> 
> 
>>Franca,
>>
>>I suggest you try 0.4.9(or current) before debugging further. There was
>>quite a few bugfixes in 0.4.9 and some of them were related to route
>>calculation. They could lead to the situation you describe if I remember
>>correctly.
>>
>>- Andreas
>>
>>Franca Delmastro wrote:
>>
>>>Hi guys!
>>>It could be probable that my processor was at 100% during those
> 
> experiments,
> 
>>>but if it was why all nodes received Hello packets from their 1-hop
>>>neighbors but they don't receive information about nodes at 2, 3 or 4
> 
> hops
> 
>>>distance???
>>>Packet collisions or exchange context between multiple threads should
>>>involve also hello packets!?
>>>
>>>Using ad hoc networks each node behaves both as client application and
> 
> as
> 
>>>router, managing  routing packets and topology changes. Building
>>>applications for multi-hop ad hoc networks I need to evaluate routing
>>>performances especially increasing number of nodes.
>>>
>>>I'm running olsr on linux kernel 2.6.9 and my configuration file is
> 
> attached
> 
>>>to this email.
>>>Another question is: how can I increase the priority of olsr process and
> 
> its
> 
>>>traffic on the network?
>>>
>>>
>>>Thanks a lot for your help!
>>>Franca
>>>
>>>
>>>------------------------------------------------------------------------
>>>
>>>_______________________________________________
>>>olsr-users mailing list
>>>(spam-protected)
>>>https://www.olsr.org/mailman/listinfo/olsr-users
>>
>>-- 
>>Andreas Tønnesen
>>http://www.olsr.org
>>_______________________________________________
>>olsr-users mailing list
>>(spam-protected)
>>https://www.olsr.org/mailman/listinfo/olsr-users
> 
> 
> _______________________________________________
> olsr-users mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-users

-- 
Andreas Tønnesen
http://www.olsr.org