[Olsr-users] Netjson bug

[Gabriel]

Il 2017-05-17 10:28 Henning Rogge ha scritto:
> On Wed, May 17, 2017 at 10:14 AM, Gabriel <(spam-protected)> 
> wrote:
>> Some weird stuff happens in the netjson:
>> Here you can see the network graphically
>> http://ninux-graph.netjson.org/topology/53b35ab9-aa6d-4705-89c3-310eac09b0da/
>> here's the netjson
>> http://ninux-graph.netjson.org/api/topology/53b35ab9-aa6d-4705-89c3-310eac09b0da/
>> 
>> The attached addresses of the node "id_172.19.185.11" are wrong:
>> id_172.19.185.11_172.19.64.1 and id_172.19.185.11_172.19.64.2 are 
>> actually
>> belonging to it's neighbour "id_10.150.9.1".
> 
> I don't think this is a bug, its one of the design decisions of OLSRv2.
> 
> The interface addresses of a node are reported by their one-hop
> neighbors in the TCs... only real "attached networks" are reported
> directly inside the TC of the local node.
> 
> I think there should be an option not to make the interface IPs
> routable at all, this might help to clean up the graph... I wonder why
> this is not done by default. What olsrd2 version are you using?
> 
> Henning

This is 10.150.9.1
(spam-protected):~# olsrd2 -v
  OLSRd2 version 0.12.4
  Git commit: v0.12.4-archive


This is 172.19.185.11

  OLSRd2 version 0.11.4
  Git commit: v0.11.4-archive


I'll try to update the latter.

Another problem that comes to my mind now is that if I need to install 
an additional plugin (like remote_control or auto_ll4) on a node I must 
compile olsr2 again and install it manually.
Would be possible to install a plugin separately like in olsrd?
Probably the makefile structure must be changed, but I think the project 
would benefit in the long run.

Gabriel