On 2014-08-20 05:29, Anton Aylward wrote:
On 08/19/2014 09:37 PM, Linda Walsh wrote:
Carlos E. R. wrote:
Right. Fetchmail *has* to hand of to something. Carlos seems to think that in the absence of a "mda" directive it hands off to sendmail/postfix. THIS IS NOT SO!
The docco says that in the absence of a mda directive, fetchmail sends to port 25 of the machine that fetchmail is running on.
And port 25, in a default configured openSUSE machine nowdays, is handled by postfix. Previously it was sendmail. Other distributions might use different things, but I don't care about them. And people may install a different MTA. That's is my meaning of "fetchmail handing over to sendmail/postfix". Not that fetchmail calls any of those programs: it does not know the name of the MTA nor cares (well, it may know, but for simplicity we'll assume it doesn't).
There is then th hope that this is properly configured to recognise the address/domain as local and will effect local delivery. This in turn _might_ end up having a .forward being used rather than delivery to the local /var/spool/mail.
In this situation I believe fetchmail doesn't forward to a user@host.domain, just to a "user", so there is no chance of sending outside by mistake. Of course, if there is a ".forward" file, that is not handled by fetchmail.
Hopefully that .forward invokes a procmail and the procmailrc has suitable defaults.
I don't remember who handles .forward files, but I think it is postfix. Normally postfix invokes procmail directly, not via .forward.
But all this is dependent on how the sendmail/postfix is configured. Any 'errors' there, any unexpected processing based on what's in the email envelope & headers and its lost.
Nay, I've never seen that. :-) And I have been using this same method for decades, even as a Linux novice. -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)