[OLPC Networking] Issues with the wireless mesh devices

[Kim Hawtin]

info at olpc-peru.info wrote:
> Some ideas:
> 
> a) Google and other intense information providers have set up their 
> servers to refuse "too many" requests from the same user in a short 
> period of time.
> b) "filtering" (nat or caching software) will be a tool to do the same 
> not only with http but with all the TCP/IP protocols.
> c) If a pair of XOs turn down an AP... it is an "internal" DOS.  But, in 
> my humble opinion, I wont call it a DOS because there is no intentional 
> attack to the server. 

perhaps you miss-read my earlier post. the atheros radios stops working and
need to be reset by a kernel module reload on most occasions or power cycle.

i expect that IP filtering would not help. i've not been able to predictably
repeat it yet.

below is the info from the network admins as to what was running on the AP[1]
running FreeBSD[2].

we have also had the same problem on some cisco APs using atheros with virtual
ESSID support at work. more info as i find it.

regards,

kim
--
[1] system: WRAP2 geode 266mhz
    AP interface: Senao 8602 PLUS (FCC) 400mw abg
    Backbone interface: Senao 8602 (ETSI) LFP 100mw abg

[2] kern.version: FreeBSD 6.2-RELEASE-p5 #1: Tue Jul 24 20:41:51 CST 2007
    hw.ath.hal.version: 0.9.17.2