[c-nsp] Weird ARP Issue with 3rd party equipment

[Mike]

Paul Stewart wrote:
> The CMM module and the Access Points never have an issue - the subscriber
> units have the problem of being reachable/pingable.  When they are not
> reachable, the web interface doesn't work neither.  All of the Motorola
> equipment involved are identical units with same software revisions.
>
>  
>   

So in *some* wireless environments, especially the canopy platform, 
bridging is handled a little bit differently than what 802.1d states 
concerning 'flooding'. The deal is that the Access points _do not flood_ 
frames with unknown mac addresses, and only build their bridging tables 
based on what they see comming from the subscribers.  This is why then 
if you clear your arp cache you get ping connectivity again - clearing 
arp means your router will send to the broadcast address in order to 
resolve the mac/ip, which then causes the access point to flood the 
broadcast, and an arp reply comes back which fills in the AP's bridging 
table and establishing connectivity.

Normally, this can 'become a problem' if your access point is rebooted. 
The mac table will be cleared on boot and it will be only until another 
arp/broadcast comes in that the table is repopulated per the above. So 
the first thing to check then is, is your access point rebooting (check 
it's uptime)? And secondly, when you are experiencing the problem, can 
you verify that the subscriber's mac address is in the bridge table? I 
betcha the answer to #1 is yes and the answer to #2 is no.

Mike-