[Xorp-hackers] Some fixes to OLSR (topology.cc)

Ben Greear greearb at candelatech.com
Mon Jun 13 13:53:21 PDT 2011


On 06/13/2011 01:01 PM, Philip Hugg wrote:
> Hello Ben /all
>
> I have attached the modified file 'topology.cc' that contains the
> changes I've made
> in the attempt to make OLSR more stable. Unfortunately, there are still
> problems
> with the OLSR but at least we are one step closer.
>
> Tested Topologies:
>
> 1) 9 nodes inline (wireless)
> n9 - n8 - n7 - n6 - n1 - n2 - n3 - n4 - n5
>
>
> 2) 16 nodes, in 4x4 grid (wireless)
>
> n1 - n2 - n3 - n4
> n5 - n6 - n7 - n8
> n9 - 10 - 11 - 12
> 13 - 14 - 15 - 16
>
> where, N1 is nei with N2, n5, n6,
> N2 is nei with n1,n3,n5,n6,n7
> N3 is nei with N2,n4,n6,n7,n8
> ...
> N16 is nei with n11, n12, n15.
>
> Topology is stable however a ping from N1 to N16 may not take the
> shortest path
> (n1-n6-n11-n16) which is 3 hops. Instead, it may follow path from
> N1-N2-N7-N12-N16 (4-hops?).
> When I re-issue the ping many times, it may take a different paths.

I'm not sure how OLSR is supposed to work exactly, but that does
sound buggy.  Could you open a bug with that test case and results?

>
> 3) Mobility test: 9 nodes inline (as above).
> When moving N9 from N8 to N7 until we get
>
> N9
> |
> N8 - N7 - N6 - N1 ...
>
> In this case, OLSR asserts.

Can you open a bug with the test case and assert message?  Might help the
next person to attempt this...

> These are some of the scenarios I used to find the bugs. However, I was
> pulled off
> this work and assigned something else. I just thought I would contribute
> my findings
> if anyone want to continue.

Thanks for the patch.

But, any chance you could generate a unified patch, ie: diff -u
or something like that?

That's a lot easier for me to read.

If you don't have time to open these bugs or generate a new diff,
let me know and I'll take care of it.

Thanks,
Ben

-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com



More information about the Xorp-hackers mailing list