Re: ppp loopback

[Carey Evans <c.evans@clear.net.nz>]

mfrattola@mail.enjoy.it writes:

[snip]

> But .. it doesn't start ip-up properly, since
> ip-up does the following:

[snip]

> # routing
> route del default 
> route add default dev ppp0
> 
> # send mail
> echo 'doing putmail' > /dev/console
> $EXECDIR/putmail
> 
> # get mail
> echo 'doing getmail' > /dev/console
> $EXECDIR/getmail

[snip]

> and I get messages from the echo lines, but I don't get any mail
> popped or sent.  Strange is, if I run ip-up by hand, it behaves
> properly.

Do putmail and getmail rely on any settings of uid?  Slackware 3.0's
pppd ran ip-up (for me) with euid=root, uid=carey.  Debian's pppd runs
ip-up with euid=root, uid=root.  I'm using the gid=carey to work out
who invoked pppd.

> Also, in debian pppd I have to set default route by hand, since pppd
> informs me that he refuses to substitute my default route (usually
> to eth0) to ppp0 I'm using defaultroute in options file, but this
> doesn't help.

Are you sure the previous default route is right?  For example, at
work we use 192.168.17.* and 192.168.18.* subnets, where
192.168.17.200 aka 192.168.18.200 is the router.  My machine is
192.168.17.11, so I can access the first subnet OK with route add -net
192.168.17.0.  To access the second I could route add default
192.168.17.200 (I think), but the better solution is route add
-net 192.168.18.0 gw 192.168.17.200.

The best solution is to start "routed -q" which updates my machine
from the router's periodic broadcasts.  I think I should put some
stuff in /etc/gateways in case someone else's machine starts
broadcasting bogus routes though.

-- 
                Carey Evans  <*>  c.evans@clear.net.nz

		 "Double, double, toil and trouble, /
		   Fire burn and cauldron bubble."