20 Sep
2014
20 Sep
'14
1:59 a.m.
On 19/09/14 22:27, Gui Iribarren wrote:
On 14/09/14 21:09, Gui Iribarren wrote:
meh... i rebooted the 4 nodes at once and things went back to normal (?)
will give it one more chance, leaving it running for a few days more.
(rebooting only labanda-o (the "phantom" batman gateway) didn't fix
things)
cheers
On 14/09/14 20:06, Pau wrote:
miaumiau, casapuente and
alfredo see a "phantom" batman gateway (labanda-o)
while labanda-o is clearly gw_mode=client, and at best "relaying"
gateway announcement from another node farther in the batman cloud
(64:70:02:4e:c5:80) I don't know any con yet :) But we should
see about the performance of both and compare.
so it seems 11s is mangling in some way the mac addresses in packets,
breaking batman-adv running on top of that layer1
root@miaumiau:~# batctl gwl
Gateway (#/255) Nexthop [outgoingIF]: advertised
uplink bandwidth ... [B.A.T.M.A.N. adv 2014.2.0, MainIF/MAC:
eth0-167/02:e6:fc:be:28:9d (bat0)]
64:70:02:4e:c5:80 (192) 64:70:02:3d:91:3c [wlan0_adhoc-167]: 10.0/2.0
MBit
=> 64:70:02:3d:91:3c (239) 64:70:02:3d:91:3c [wlan0_adhoc-167]: 10.0/2.0
MBit
palmeras_eth0-167 (168) 90:f6:52:c6:00:f0 [wlan0_adhoc-167]: 10.0/2.0
MBit
too bad, i'm dismantling the test setup...
anyway, from informal tests running netperf before and after, i didn't
see a significant difference in throughput or stability, between adhoc
and 11s mode
so, it looked very promising but maybe it's not quite "there" yet
the good news is that ath9k adhoc mode is looking astonishingly stable
so far, been running for two weeks with no hangups... Should we send
Felix some chocolates? or maybe cut his hands to keep the code intact?
cheers!
gui
very nice pau and henning for the tip
cheers!
gui
_______________________________________________
Dev mailing list
Dev(a)lists.libre-mesh.org
https://lists.libre-mesh.org/mailman/listinfo/dev
The layer1 (11s) links you with the direct neighbors, the layer2 is batman-adv but only
for your local cloud or neighborhood, then the layer3 (bmx6) attach all the clouds in a
single big routed network.
On 14/09/14 01:59, Charles N Wyble wrote:
_______________________________________________
Dev mailing list
Dev(a)lists.libre-mesh.org
https://lists.libre-mesh.org/mailman/listinfo/dev
On 9/13/2014 5:48 AM, Pau wrote:
> Hi.
>
>
>
> Using 11s instead of adhoc for deploying our mesh networks can bring
> some interesting features, among others:
>
> 1. Better support for 11n
> 2. Better compatibility with drivers (probably even ath_htc works fine)
> 3. You can bridge it to another interface if necessary.
> 4. It does NOT try to synchronize the TSF counter of your wifi card thus
> you can create up to 8 11s VAP mixed with adhoc, AP, client, etc...
> 5. I don't know deeply 11s but it probably has better design for
> deploying mesh networks
I like the pros.
What are the cons?
>
> I've been testing it in a 12 nodes mesh network mixing it with bmx6 and
> it worked as a charm. No more "strange" problems coming from the adhoc
> layer.
Very nice.
>
> Yesterday Gio the great added support for 11s in libre-mesh [1]. So now
> we only need to test it a bit more with our already crazy network
> architecture, so it may became even a bit more crazy for traditional
> mesh folks (bmx6+bat-adv+11s, never done something like that!).
So 11s would be layer1, bat-adv l2, bmx6 l3 ?
_______________________________________________
Dev mailing list
Dev(a)lists.libre-mesh.org
https://lists.libre-mesh.org/mailman/listinfo/dev