On Mon, 30 Apr 2001, Wayne Storey wrote:
> We have a situation where one of our customers is using Exim for sending
> mail and on occasion it seems to not be sending email out.
You posted this message 3 times to the exim-users list, from an eastern
time zone:
Date: Mon, 30 Apr 2001 13:36:25 +1000
If the repetition was a mistake, fine, we all make mistakes. If you were
trying to get a faster answer, please don't repeat. It doesn't help.
Indeed, I always go for the last posting, so repetition in fact delays any
response I might make.
Also note that most Exim users are in more western time zones, so you
are likely to see delays. Here, for example, it is less than an hour
since the start of my working week.
> The first time I noticed this there was about 3 Frozen emails at the top of
> the queue and 30 non-frozen. I deleted the 3 frozen and the rest started to
> be sent.
That is odd. Frozen mails should simply be skipped over. They should not
block the sending of others. Indeed, Exim normally process the queue in
a random order (unless you set queue_run_in_order).
> Now this has happened again, this time the customer says that there were no
> frozen emails but when he forced the top email back to the sender the rest
> of the queue started to go through ok.
I think we need more specific information in order to figure out what
was actually happening. If it happens again, save a listing of the queue
(exim -bp) and also save Exim's log for some time before and after,
preferable enough of it to see all the entries for the frozen messages
and the others on the queue to see why they are not being processed.
> My question is, is there a way to delete mail automatically if it sits in
> the mail queue, say 4 or 8 hours and send it back to the user asking them to
> resend their message, either by a script or within Exim.
The retry rules are provided to control retrying. You can also make use
of $message_age if you want to route a message differently after some
time has passed. However, what is the point of asking a user to resend
if the message is failing to be delivered? A resent message is likely to
fail in the same way.
> Also why would one or two emails stop the rest of the emails going through.
They should not. If you can generate some evidence of a situation where
this happens, please send it to me. For example, output from a queue run
with debugging (exim -d9 -q) would be useful. The debugging output is
written to stderr, so you can catch it in a file with
exim -d9 -q 2>/some/file
If the queue is long, the output might be quite voluminous.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
