[Testing] Questions about mesh path discovery

[Michail Bletsas]

testing-bounces at lists.laptop.org wrote on 07/02/2007 03:13:22 AM:

> Hi :
>   I’m Bryan from Quanta, we are doing some testing about mesh network,
> I have two questions, if you know how to resolve, please giving 
> instructions, thanks !
> 
> OFW : Q2C17
> Image Version: 406.14
> Machine: B3
> 
> 1) we did a multi-MPP experiment, inspired from the link: http:
> //wiki.laptop.org/go/Establishing_a_Mesh_Portal.
> Configure two MPPs (MPP1 and MPP2, only one DHCP server on MPP1)
> and three MPs, we found all three MPs establish their connection to 
MPP1.
> MPP1 and MPP2 are physically very close( < 0.5M), so it maybe 
> shouldn’t influenced by different distance.
> Is there any manual way could assign the connected MPP?

Bryan,

For that test to work you have to use the wireless firmware's blinding 
table feature.
Please check http://wiki.laptop.org/go/Wireless_Driver_README for info on 
how to manipulate the blinding table.


> 
> 2) Having another test for discovering communication path. 
> Arrange a topology that MP2 connect to a infrastructure AP via MP1 
> and MPP1, like below shows:
>   AP <------------MPP1<-------L1--------MP1<--------L2---------MP2,
>   The distance between MPP1 and MP1 is L1 when L2 is from MP1 to MP2.
>   L1 and L2 are about equal 100Ms, so we assume MP2 should connect 
> MP1 first before arriving MPP1.
>   But when I issued “tracepath”, I just saw the path as “MP2àMPP1à
> AP”, MP1 wasn’t appear on the routing list,
>   Is anything wrong ?

OLPC's mesh network operates at layer-2 (at the ethernet frame layer).
Tracepath is a leyer-3 tool and won't reveal the layer-2 path through the 
mesh clound
You have to use the forwarding table inspection commands to find the 
layer-2 path.





Best,

M.