0.2.4 deatuhentication reason_code=3

[Ondrej Zima]

Kelly Hogan wrote:
> Jouni,
> 
> We recently upgraded to 0.2.4 on soekris with a dual radio
> configuration.  On our master radio, we are getting deauthentication
> reason_code=3 from nodes with signals of -93/-70 which seems odd.
> 
> They then never re-authenticate unless we reboot the client side, then
> they last for only a few minutes.  
> 
> Ideas?  Should we downgrade to 0.1.3?

Hi,
I discover some stupid behaviour. It is not validated, but I have the same sitation. The only solution I make is the reset the card if client not response (in cron I have the script with checking the associated clients and if they not reply to ping, iwpriw reset 0 repair it) It wors, but not great, coz there was delay to run the cron script (was start every fitfth minute).
Now, I try to play with ALC and discover this: after the card is reset, the ALC is set to 1 (ALC is enabled) TX power fluctuate (MIF register 62) and driver will ignore my txpower changes. If I set the ALC to 0 (disabled) and set my txpower settings, it works very good. But only to next card reset, which can appear frequently (it depends on client quality).
For this, I create simple script which check the MIF 62 and test it for my txpower settings. If it vary, it seems the card was reseted and ALC is enabled. In this situation I disable ALC and reset the txpower.
If you use the patch from jbohac, it internaly disables the ALC if card was reseted.

BTW: deathentication for reason_code=3 is deauthentication for client inactivity



-- 
Ondrej Zima <amiandrew at volny.cz>
Member of Czech ATO - Amiga Translators Organization