Philip Hazel wrote:
> > if a user ivan on say, horse, has a .forward file containing:
> >
> > someone@???
> > \ivan
> >
> > he expects to receive a copy of incoming mail at his mailbox on horse and
> > another one copy sent to someone@???. But instead the second line
> > of his .forward generates ivan@??? and the message is passed to
> > tiger, where it is looked up in forwardtable again, sent back to horse ad
> > infinitum. It seems that check_ancestor doesn't work when qualify_domain
> > are not recognized as local domains. Looks like a bug to me.
>
> How is the message forwarded from tiger to horse? Do you transform the
> address into ivan@horse? In that case, the incoming address is
> ivan@horse and so of course check_ancestor won't match when checking it
> againsg ivan@???.
>
> > Tiger also has director forwardtable
> > forwardtable:
> > driver = aliasfile;
> > file = forwardtable,
> > search_type = nis,
>
> Does this turn ivan@??? into ivan@???? I'm guessing
> it does, leading to check_ancestor not working as you expect.
>
> Have you tried setting
>
> qualify_recipient = horse.jet.msk.su
>
> on horse? Then any unqualified recipient addresses on that system will
> be treated as truly local, but since you have got
>
> qualify_domain = jet.msk.su
>
> the *sender* addresses will look like ivan@???.
>
This would break things. Imagine user a receives mail on horse, user b
receives mail on hound. If I set quailfy_recipient to horse.jet.msk.su
attempt to mail to unqualified user b from horse (for example from
a's .forward file will perform local delivery on horse and user b will
never receive this message (I forgot to mention that horse and hound have
synchronized passwds).
I think I'll prohibit use of backslash in .forward files using something
like "real-" prefix instead. But I'll be grateful if someone can think of a
way to preserve backslash functionality.
avk
--
--
"An NT server can be run by an idiot, and usually is." -- Tom Holub
--
*** Exim information can be found at
http://www.exim.org/ ***
