[Battlemesh] Network configuration and address plan for Battlemesh v9

[Bastian Bittorf]

* Henning Rogge <hrogge at gmail.com> [12.08.2015 14:31]:
> > i like this idea, i just was just to shy for proposing it.
> 
> I think a /24 for each router should be enough... we could add the
> whole /24 on the external LAN, but just (as proposed above) only give
> out a part of the addresses via DHCP.

its not about the DHCP space, but keeping in mind (for the config)
that we can have more then 2 radio's.

> > what also circles in my head:
> > for now we dont use any traffic-prioritizing of the signaling/mesh-proto
> > traffic, dont we?
> 
> OLSRd/OLSRd2 both set the TC flags to get "high priority"... which
> should (I think) be handled both by the default queuing discipline and
> by the wifi driver.

no, does not work. it's easy to kick away a route whith saturating a
link. here in our community we have found this issue e.g. on a weak
5ghz link (3mbit only). when making traffic the ETX-values are getting
bad and more bad >ETX 20 and the route switches (needs <30 sec)...

marking the traffic with DSCP CS6 or CS7 is ok, but not enough.

here the part of the source in olsr1:
http://www.olsr.org/git/?p=olsrd.git;a=blob;f=src/unix/ifnet.c;h=252be247a44365fc77ee7a2813bea5aff5caf2de;hb=HEAD#l778

and olsrd2:
http://www.olsr.org/git/?p=oonf.git;a=blob;f=src-plugins/subsystems/oonf_packet_socket.c;h=ffc442cbbde306c6192092858c0ff72342479916;hb=HEAD#l663

i can see that it applys correctly with:
user at laptop:~$ sudo tcpdump -v -n ip and ip[1]!=0 -i wlan0
tcpdump: listening on wlan0, link-type EN10MB (Ethernet), capture size 65535 bytes
14:50:22.289593 IP (tos 0xc0, ttl 1, id 0, offset 0, flags [DF], proto IGMP (2), length 32, options (RA)) 0.0.0.0 > 224.0.0.1: igmp query v2

so TOS is '0xc0' which corresponds to DSCP/CS6 = voice

how are other protos are handling this?

bye, bastian