[Olsr-users] OLSR routing issue with highly connected node

[Mitar]

 Hi!

Nodes connect to both VPN servers. And then they use olsr to route
over one or another connection.

So this is one VPN server (10.254.0.4):

https://nodes.wlan-si.net/node/vpn-b/

And another (10.254.0.2):

https://nodes.wlan-si.net/node/implicator/

It seems node gets routes only from one VPN server, the one with less
peers itself. And not from another. But for some reason it does not
get routes from the smaller VPN server even for two-hop away nodes.

Is it possible that because it is better to talk to nodes connected
directly to 10.254.0.2 over 10.254.0.2, node does not add routes for
nodes connected directly to 10.254.0.2, but for some reason also nodes
by 10.254.0.2 are not added?


Mitar

On Mon, Dec 22, 2014 at 8:57 AM, Mitar <(spam-protected)> wrote:
> Hi!
>
> So most of the time the olsr routing table at the node with low MTU is
> empty. Sometimes it gets filled for a bit, but even then partially.
>
> With older version of olsrd the moment when it was a bit filled was
> looking like:
>
> http://pastebin.com/RgRcvAu4
>
> With the latest version of olsrd both on the server and client it
> looked a bit fuller, but still now complete:
>
> http://pastebin.com/8uJMEQh5
>
> And then it got emptied. It is missing obvious routes like default
> route. When routing table is non-empty it seems it has nodes only to
> those nodes which are directly connected to the same VPN server.
> (10.254.0.4 in the log above.) Gateway is further away, so that's why
> there is no default route.
>
>
> Mitar
>
> On Mon, Dec 22, 2014 at 8:46 AM, Henning Rogge <(spam-protected)> wrote:
>> olsrd already looks at the MTU value of an interface and should not
>> send packets large enough that they must be fragmented. If that
>> doesn't work its a bug.
>>
>> Henning
>>
>> On Mon, Dec 22, 2014 at 8:44 AM, Teco Boot <(spam-protected)> wrote:
>>> Can we unset the DF flag, so olsrd can use IP fragmentation if something goes wrong keeping UDP packets <= MTU?
>>>
>>> Teco
>>>
>>>
>>>> Op 22 dec. 2014, om 08:33 heeft Henning Rogge <(spam-protected)> het volgende geschreven:
>>>>
>>>> Can you do a look into the txtinfo output of the node?
>>>>
>>>> Is the 1-hop/2-hop neighbor table okay?
>>>>
>>>> I assume that the problem is somewhere in the Hello
>>>> generation/processing but I would like to be sure before going onto a
>>>> wild hunt.
>>>>
>>>> Henning Rogge
>>>>
>>>> On Mon, Dec 22, 2014 at 8:31 AM, Jernej Kos <(spam-protected)> wrote:
>>>>> Hello!
>>>>>
>>>>> On 21. 12. 2014 22:04, Henning Rogge wrote:
>>>>>> So you are using the normal (unmodified) olsrd code and just have
>>>>>> links which might (or might not) have a lower MTU?
>>>>>
>>>>> We are using the unmodified olsrd code, version 0.6.7.1. VPN links have
>>>>> different MTU values because our VPN tunnels perform PMTU discovery and
>>>>> adapt the tunnel MTU accordingly.
>>>>>
>>>>> To handle this on the server side, we use one bridge per MTU value (so
>>>>> VPN clients with the same MTU are all on the same bridge). And then we
>>>>> have all the bridges configured in olsrd as interfaces.
>>>>>
>>>>>
>>>>> Jernej
>>>>>
>>>>
>>>> --
>>>> Olsr-users mailing list
>>>> (spam-protected)
>>>> https://lists.olsr.org/mailman/listinfo/olsr-users
>>>
>>
>> --
>> Olsr-users mailing list
>> (spam-protected)
>> https://lists.olsr.org/mailman/listinfo/olsr-users
>
>
>
> --
> http://mitar.tnode.com/
> https://twitter.com/mitar_m



-- 
http://mitar.tnode.com/
https://twitter.com/mitar_m