On Thu, 30 Jun 2005, Ed W wrote:
Seemed like a good solution to me because the point was I wanted a way
to bypass all my usual postfix restrictions.
Set up the "receiving" side of a content filter on, say, port 10025, and
redirect fetchmail there. OTOH, usually 127.0.0.1 would be in Postfix's
mynetworks, and "permit_mynetworks" early in the restrictions, or
restrictions that don't apply to 127.0.0.1, would fix your problem while
still using the SMTP port (BTW, Postfix's SMTP port is faster than
Postfix's sendmail interface).
Situation is that I
normally reject mail which looks "suspicious" like having an email
address which is obviously fake. At least this way someone might get a
bounce or failure message if it's really genuine. However, by the time
fetchmail is handling mail it's going to become a bounce, so I am in the
awkward position of effectively silently discarding users' mail (because
the orig email address appears to be fake or at least it's unreplyable).
That is something you cannot fix with any setup, and you can IIRC defeat
bounces altogether.
Also, since changing back to using the SMTP delivery option I get
occasional points where fetchmail dies with "terminated with signal
13". I'm going to set it up under daemontools if I can figure out the
correct fetchmail incantations...
daemontools isn't going to help, and buys you into proprietary,
time-limited software. A free (GPL or something) and halfway compatible
replacement is "runit", somewhere on http://smarden.org/
--
Matthias Andree