On Fri, 14 Jul 2000, Bernard Stern wrote:
> news_bounce:
> driver = lookuphost
> domains = *
> local_parts = news
> condition = ${if and {{> {$message_age}{43200}}{eq {$sender_address}{""}}} {yes}{no}}
> # 43200 s = 12 h
> transport = trash
>
> This works well, but apparently only for domains that have an
> MX record. This does not work for other domains (mmhhh... I wrote
> apparently because I did some checks but I did not check every
> instance of such hosts from my log files :-) ).
I wonder... there are domains out there that don't have MX records, but
whose zones are on broken name servers so that when you try to look up
the MX record, you get a name server error instead of "no such domain".
If this happens, Exim cannot proceed; it has to defer delivery and try
again later. That would defeat this scheme.
> As I understand the spec, MX records then A records are looked up,
> so this router should work for both types of domains.
Indeed it should.
> While
> MX-domain bounces vanish as expected after 12+ hours, A-domain
> bounces keep stuck in the "queue".
Is this *all* A-domains, or just some. I predict "some", and if you look
for their MX domains you get a temporary DNS error.
> Looking at the spec, there is the mx_domains options, but as
> I understand the spec, this would not meet my needs.
No.
> So my question: is there an inherent difference in the treatment
> of hosts with and without MX record using only these options?
There should not be. Have a look at the log to see what the entries for
these stuck messages say. You could also try delivering one with -d9 set
to see what the debugging output is.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
