RE: Strange bind/sendmail issue
Sorry for the long lines, I know about them...
I'll move the lists stuff over to a different MUA some time. :)
> Take a look at /etc/resolv.conf and look at the search path.
> For a mail
> server, you may want to completely remove the search line.
Exactly my first thought, but the search directive has only the internal NS zone defined, "customer.lan" and not "customer.com".
>
> On Wed, 27 Sep 2006, Erik Dörnbach wrote:
>
> > Hello list,
> >
> > at one customer's site I encountered a really strange
> issue, regarding sendmails handling of emails with
> non-existing domains as recipient... running bind9 and sendmail.
> >
> > In detail, for what ever reason a mail with non-existing
> domain may enter the sendmail queue, it appears with the
> customer's own domain trailed to the non-existing one, making
> it appear as if it was a subdomain, likewise it's treated as one.
> >
> > For example:
> >
> > - Customer has domain "customer.com", the server in
> question is the mx for that domain
> > - Mail enters the queue with recipient address
> "user@doesnotexist.net"
> > - Mail is listed in queue/log as
> "user@doesnotexist.net.customer.com"
> >
> > What is going on? I don't have this phenomen on any other
> of my machines with quite the same configuration running.
> It's not much of a problem, but I would like to understand
> what is going on there, either on the resolver or sendmail.
> And have a cleaner mail queue of course! ;)
> >
> > I can give more information about sendmail config, bind
> config, ns queries and so on as needed on request.
> >
> > The only thing that caught my attention, the authoritive NS
> for "customer.com" (not under my control) does some special
> magic to make all queries to non-existing hosts under
> "customer.com" point to itself. I'm not sure what that is
> for, but that's why sendmail tries to deliever those emails
> to that NS.
> >
> > But then this can't be the source of the problem... I think.
> >
> >
> > THX and with best regards,
> >
> > Erik
> >
> >
>
Reply to: