Chun-Yeow,
yes, you are correct an extra node should not affect the rest of the 
nodes in the mesh. Adjusting the "lifetime" is NOT the solution (it does 
add overhead).

--Fabrice

On 12/8/2011 8:43 AM, Yeoh Chun-Yeow wrote:
> Hi, Fabrice
>
> From README authsae:
> lifetime: how long a negotiated PMK is valid for. At then end of this 
> period the key will be deleted and the peer will be authenticated 
> again (if it's still around).
>
> I think that it could be a problem whenever one extra mesh node is up 
> in your network, the communication of existing mesh nodes are 
> affected. So the only way to minimize the problem is to reduce the 
> lifetime, but then it will create extra overhead in the network.
>
> Regards,
> Chun-Yeow
>
> On Thu, Dec 8, 2011 at 9:39 PM, Fabrice Deyber 
> <[email protected] <mailto:[email protected]>> wrote:
>
>     Hi Chun-Yeow,
>     Yes, that 's what I observe. It does go back to normal after some
>     time. That time is related to the "lifetime" in the SAE config file.
>     This makes sense, the keys are regenerated every "lifetime". It
>     also correlates with the fact that restarting the meshd daemon(s)
>     fixes the problem.
>
>     --Fabrice
>
>
>
>     On 12/8/2011 5:03 AM, Yeoh Chun-Yeow wrote:
>
>         Hi, Javier and Fabrice
>
>         After debugging with Wireshark, let me clarify further:
>
>         Since Node 1 is pinging PC behind Node 2, after successfully
>         ping and node 3 is brought up, if the ping is continue, PREQ
>         broadcasting by Node 1 cannot be received or decrypted by Node
>         2. Thus, Node 1 assumes Node 2 is dead or out of range since
>         no PREP from Node 2. Ping is route through Node 3.
>
>         It seems that the problem is the broadcast PREQ frame
>         generated by either Node 2 or Node 1 cannot be decoded by Node
>         1 or Node 2 respectively whenever Node 3 is up or reboot. Only
>         node 3 can decode the broadcast PREQ frame from either Node 1
>         or Node 2.
>
>         But after a period of time, it backs to normal.
>
>         Regards,
>         Chun-Yeow
>
>
_______________________________________________
Devel mailing list
[email protected]
http://open80211s.com/mailman/listinfo/devel

Reply via email to