Currently, I have a problem with pppd where the line drops, and pppd
doesn't get a HUP.  As a result init never spawns a new one.  I can debug
this once I'm physcially at the machine, but I'm not (I'm in CA, it's in PA),
And keeping this link alive is important, but rebooting every 2 hours
is not an answer I like (which is the solution I came up with)...

At any rate, I've written two scripts which should detect when the conneciton
gets stale.  One uses ping, and the other watches the syslog file getting
pppd's LCP echos.  Both work as they should when I run them as a user.  
However, if I have pppd start them from ip-up, they are out there and
running, but they never do any work.  They both follow the paradigm:

while /usr/bin/true; do
	sleep 600
	check if ppp0 is alive
	if dead kill pppd and exit
done

nothing special about it.  All scripts are Bourne shell scripts.  I've had
the script write to both the syslog (via logger) and to a file as well.  At
first I thought logger was behaving strangely, but writing to a file
confirmed it was just the whole script :-(

I've tried everything from nohuping the scripts to incorporating them
directly into ip-up.  No change in behavior tho.  If the scripts were
exiting right away, I'd understand why they do no work :-)  However, I'm
somewhat confuzzeled when I see the script out there and it's marked as
I+, which implies it's doing some amount of work.  Neither file shows
any thing being written tho.

I'm out of ideas.  Anyone ever see this kind of behavior before?

Also what kind of ftp speeds are people seeing out there with 28.8Kb
modems?  I'm getting 3.7Kb/s from the machine, bit 1.2Kb/s to the machine.
That seems really odd.  Why would the speed be so much slower going to
the machine for the same file?  Anyone else see this kind of thing?  I've
got the DTE locked in at 57600 and the DCE typically reports 21600 or so...