On Tue, 23 Oct 2001, Dan Egli wrote:
> Ok. I've got a question. On one of the servers I'm administrating, the mail
> system has SEVERAL messages over two weeks old that are undeliverable, but
> have not expired or frozen, despite my settings. Observe:
>
> 20d 49K 15ovQR-00071Z-00 <(address removed)>
> ray.warner@???
>
> 18d 3.0K 15pWYK-0000Ts-00 <(address removed)>
> islandhome@???
>
> 17d 3.6K 15px93-0001pL-00 <(address removed)>
> islandhome@???
>
> 17, 18, and 20 days? Come on! Here's my retry config:
>
> * * F,2h,15m; G,16h,1h,1.5; F,4d,8h
> Now correct me if I'm wrong, but that translates to:
> Try every 15 mins till message is 2 hours old, then
> Then Try an hour, and then gain 1.5 hours till the message is 16 hours old,
> then
> try the message evry 8 hours till it's 4 days old.
>
> Correct?
Correct. Try running this test
exim -d9 -q 15px93-0001pL-00
and see what the debugging output says. If you can't make sense of it,
send it to me. Cancel that - no need - see below...
> The mailq routnie (exim -bp) lists these as active messages, not frozen,
> despite the problems reaching the mail servers they belong to. I read the
> logs, and it's getting DNS timeouts looking up the MX addresses for those
> sites. Observe:
>
> 2001-10-23 11:20:11 15ovQR-00071Z-00 == ray.warner@???
> R=lookuphost defer (-1): host lookup did not complete
> 2001-10-23 11:20:11 15ovQR-00071Z-00 failed to open DB file
> /var/spool/exim/db/retry: File exists
AHA!!! There's your problem. Exim can't open its retry database. So it
can't record any retry information. So the whole retry logic will be
stuffed.
Seems a really weird error though, since it should be opening the file
for update.
You can try rm /var/spool/exim/db/* to kill all the retry data, and let
Exim re-create it.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
