I recently found that that hotspot profile “keepalive-timeout” feature (default value of two minutes) uses ping to verify that a client is still alive.
If a client is running a personal firewall that blocks incoming pings (XP service pack 2, Norton, etc…), and has not sent any network traffic in the last two minutes, they will be disconnected from the hotspot and forced to re-authenticate.
Very annoying
Maybe this is old news but I couldn’t find it anywhere in the MT documentation.
ARP method is already implemented. It is used by default. I just tested - it
works for me.
Only if ARP fails ICMP is used. ARP will fail if hotspot client resides behind
a router.
To debug keepalive-timeout issues easier:
set small keepalive-timeout (6s);
start ethernet sniffer;
connect client (for hotspot to start keepalive-timeout probes);
If some bug is suspected
get sniffer output where keepalive-probes (including first keepalive
probes to that client!) are visible;
create supout file;
send both files to support with detailed problem (and possible bug)
description.
Yes, I am using the hotspot in front of a routed network. I’m sure several others are as well. Maybe the default keepalive timeout could be changed to 10 to 30 minutes or some note put in the documentation to prevent problems with future users.