Chargeint is a way to reduce your costs when you have charges based on your time online, and the interval between two charges (the Charge Interval) is relatively large (e.g. per minute).
Chargeint only hangs up two seconds before the end of a charge unit. isdnlog can be used to set the length of the charge unit (i.e. Charge Interval) according to the time of day and the date.
You can set the length of a charge unit manually via the isdnctrl parameter
chargeset, or set up isdnlog to do this automatically for you:
/sbin/isdnctrl huptimeout ippp0 5
It's best to synchronize the clock in your own computer with that of the
switching station by calling isdnlog with option
Chargeint will only hangup if there was no activity on the line. Possibly your
service provider uses a router (e.g. Cisco) which sends a "keep
alive" packets every ten seconds. If the Cisco doesn't get an answer for
its keep alive packets then it will stop routing. This normally happens after
the 4. or 5. keep alive packet. Very recently (begin of 2001), support for
Cisco's keep alive packages has been corrected, so you can either use it,
or tell the provider not to use keep alive packets
"no keepalive" in the Cisco configuration).
It could also be that it's not the keep alive packets that are keeping the connection open, but rather OSPF routing updates. The sending of these updates can only be switched off on the Cisco. You can configure "snapshot server" on the BRI interface. That means it will send out routing updates only when they are received through this interface.
However, nowadays the most likely cause for open connection is that
connection requests looking for a backdoor or a file sharing application
cause issues like this. You can use the
of ipppd to indicate which packets should be regarded as link activity.
See the man page for more details. A configuration could be like this:
active-filter 'outbound and not icmp == 3 and not tcp & 4 != 0'