Piece of cake, just add another key word to the filters, keepalive. If
the packet matches the keepalive filter (as it matches the permit and
deny) then the idle timer is reset, if it doesn't match, then the idle
timer is not reset. I've seen several ppp daemons (running on hosts)
that do this, but have never seen it implemented in a terminal
server....but I would definitely love to see Livingston do this.
>> At present, if you have a PPP connection up
>> and running, all it takes is a single ping occasionally to keep the
>> connection alive.
>Less frequently than pings, I see a POP connection every N-1 minutes
>when I have the idle set to N minutes, or an IRC session, which will
>generate continuous traffic with no input from the user. Or a telnet
>session to a shell account running IRC. Or what about auto refresh web
>pages?
Indeed, POP would be another one that would be configured to not reset
the idle timer. IRC is the one that we haven't found any way to
prevent, and auto-refreshing web pages are still rare enough that they
aren't a problem. Besides, if I were trying to defeat an idle timeout,
I think I could find a more bandwidth-friendly way to do it than
reloading web pages. :)
>While you could ignore POP and ICMP traffic when determining idleness,
>you can't ignore IRC or telnet or WWW traffic. And if you can't come
>up with a solution that works, there's probably no point in
>implementing something that users will be able to easily defeat.
Well, it works a heck of a lot better than what is available right now.
No, its not a perfect solution, but its light-years ahead of the
bit-based idle timeouts implemented right now, where even line noise
could keep the connection open!
Jeff McAdams | "A strange game.
IgLou Internet Services | The only winning move is not to play."
e-mail: jeffm@iglou.com | -- Joshua