Hello I continued with testing: Raspberry Pi b+ with a LOGILINK WL0084B USB Stick on Jessie with Kernel 4.1.13+, batman2015-0
I tested:
Batman + Ad-Hoc Batman + 802.11s only 802.11s
against each other. This lead to the following results:
All of the above configurations show basically the same behaviour: When pinging, the reply times from directly connected nodes are really stable (1-4ms, depends on the range). When pinging via one hop in between, the network seems to take some breaks. A burst of pings (5-10) are replied pretty accurate, but always followed by a really long break of about 2-3 seconds
The problem is less intense when using the batman configurations (->actually does rout better (: ).
Facing this results I think I might be staring at a driver's issue, but still wanted to see if anyone here can help me with ideas.
Thanks everyone for the time and help!
Richard
On 18 December 2015 at 18:30, Sven Eckelmann sven@narfation.org wrote:
On Friday 18 December 2015 18:22:40 no name wrote:
I made a little table were I tried wheezy against jessy and 802.11 as someone suggested earlier in the mailing list to me. I placed the raspberries next to each other on my table and tested through different settings:
ad-hoc+batman-2014.X@debian wheezy: 1m distance .. 200 pings, 0% paket loss, reply time -> (min,avg,max,mdev 1.2ms, 1.8ms, 6.4ms, 0.8ms) ad-hoc+batman-2015.2@debian jessie: 1m distance .. 200 pings, 6% paket loss, reply time -> (min,avg,max,mdev 1.1ms, 170ms, 1400ms, 370ms) 802.11s @debian jessie: 1m distance .. 200 pings, 0% paket loss, reply time -> (min,avg,max,mdev 1.05ms, 1.8ms, 14ms, 1.5ms)
Your "table" is incomplete. adhoc without batman-adv and meshpoint (802.11s without fwding+ttl=1)+batman-adv are missing. You also mixed two different batman-adv versions. Either you use the same batman-adv version everywhere or make sure that the kernel+wifi driver doesn't change. Changing multiple variables between two tests tests make the results more or less useless for comparisons.
Kind regards, Sven