Another comment on your scenario:
Each node should always be reachable (if a suitable path in the mesh exists)
via the IP of the first configured interface ( eth0 if launched as: batmand
eth0 ath0 ).
Some more background:
If batman is configured to serve multiple interfaces only one of them (the
first defined one) is announced with a TTL=50. The second and further
interfaces are by default announced with a TTL=2 only. The reason for that
is, that one known IP-address is sufficient for distant nodes to reach that
node and therefore at least one of the configured interfaces and
corresponding IP must be announced with a high (50) TTL.
Announcing all interfaces with a high TTL would only cause increased routing
tables and a waste of bandwidth due to additional OGM messages traversing the
However, all configured interfaces need to announce their existence at least
to their neighbors in order to indicate the existence of a link so that they
can be used (thats why the remaining interfaces are configured with a
Because all IPs are announced (with seperate OGMs) via all interfaces, it does
not matter which interfaces have a good connectivity to its' neighbors and
That is one interesting thing with batman: A node does not know the exact path
to a destination node, it only knows the best neighbor towards the
destination node. Therefore, in a multi-hop-scenario, you can even configure
intermediate nodes with a TTL=2 for all interfaces, making such nodes fully
invisible for all more-than-two-hops-away-nodes. Such invisible intermediate
nodes would still be used for routing end2end traffic.
On Wednesday, 14. February 2007 13:37, Benjamin Henrion wrote:
We met with several people in Brussels active on the reseaucitoyen
project to test Batman.
Our configuration was 4 laptops like this:
laptop1 <----ethernet----> laptop2
laptop3 <----ethernet----> laptop4
With 8 different IPs for the 8 interfaces.
We launched batman-0.10 on all the nodes, but it appeared that one of
the interfaces was not reachable via ping. And the experience was
reproductible when we restarted the dameon, sometimes with a different
We are going to test the same configuration soon to see if batman
chooses the fastest route in terms of throughput.
Benjamin Henrion <bhenrion at ffii.org>
FFII Brussels - +32-484-566109 - +32-2-4148403
B.A.T.M.A.N mailing list