Postfix expert needed for dialup config

[Tommy Trussell]

On Thu, 16 Dec 2004 00:20:43 +0100, Wander Boessenkool
<wander at tomaatnet.nl> wrote:
> On Wed, 2004-12-15 at 16:52 -0600, Tommy Trussell wrote:
> > On Wed, 15 Dec 2004 17:31:18 -0500, cpinto <ulist at gs1.ubuntuforums.org> wrote:
> > > Have you tried adding the command "postqueue -f" to the /etc/ppp/ip-up
> > > script?
> > > The ip-up script is executed whenever the PPP connection is successful
> > > so it might be a good idea to flush the queue at that time.
> >
> > Thank you for the suggestion, though I would prefer NOT to flush the
> > queue when the connection comes up, because it would flush all the
> > pending messages without sending them.
> >
> 
> Flushing is _not_ deleting. The postqueue -f commands will actually
> attempt to send everything that's still in the queue. Just as flushing a
> toilet, you're trying to send it down the pipe ;-)

Hmm -- the one time I tried that command, I believed the messages were
flushed completely away, but I will try again.

The command I use is sendmail -q, which is what all the docs I have
seen recommend, but if postqueue -f is better for some reason it's
certainly easy to change.
 
> > I think the actual problem I'm having may be related to how postfix
> > handles DNS -- the error message was that there was no route to the
> > host.
> 
> No route isn't a dns problem, it means it has resolved DNS, but the
> ip-stack couldn't find a route to resolved address
...
> Isn't easier to use the smtp-server of the provider in evolution? (And
> it's offline function)?

Easier for whom?  -- If I can get what should be an "ordinary" setup
working, then I don't have to explain the online/offline function in
Evolution, nor its interaction with the online and offline status of
the ppp connection and that you can only send mail when things are
right, as well as explain the error messages that come up when you get
the order wrong.

Now, if there was some way to synchronize the online / offline feature
in Evolution with the online / offline status of PPP, then I think
that would be reasonable.

Thanks for your ideas -- I will have to figure out the implication of
"the ip-stack couldn't find a route to resolved address." This
mystifies me because I could ping the server from a terminal window,
and I could connect to sites using a web browser.

Obviously the situation for postfix is not quite the same as for the
terminal window or the web browser... I was focused on DNS partly
because one of the Red Hat documents recommended using this postfix
setting

disable_dns_lookups = yes

though it said if you use this setting you should use an ip address to
specify the relayhost. To me that doesn't sound like a good solution
either....