On Sat, 26 Oct 2002, Brian Bennett wrote:
> --
> Please help me.
>
> I have a mail server that is running completely out of control.
> I currently have 2476 messages queued to be sent, that is up by
> about 200 from this morning. I currently have 193 exim processes
> doing...something.
exiwhat will tell you what they do exactly, but 193 are too much for a
usual queue-run, until you raised the maximum queue-runners.
> My problem is that these messages are undeliverable and have been
> sitting in the queue for up to 25 days. I've been trying my damnedest
> to clear out my mail spool, but I just can't seem to do it. At first
> I was having a problem with users with very full mailboxes (some are
> as much as 65M). In order to attempt to help the situation I set a
> quota limit of 5M hoping that future messages would get rejected.
25 days is too much, are those messages all frozen?
try to set these options (exim4)
ignore_errmsg_errors_after = 3d <--- this deletes undeliverable bounces
after 3 days
timeout_frozen_after = 7d <--- this removes frozen messages older
than 7 days from your queue, i'm not
sure if they are bounced or removed
split_spool_directory <--- Reduces the load a little bit
auto_thaw = 12h <--- to retry frozen messages after 12h
> Instead those messages generate errors and cause the origonal message
> to sit in the queue attempting re-delivery. The bounce message also
Please show us some log-entries and/or relevant parts of a debug-run.
> sits in the queue because the return e-mail address is often
> undeliverable as well.
you should use sender and receiver verification to deny messages to/from
undeliverable addresses.
> What I want is this:
> 1. Attempt delivery, if undeliverable, attempt retries for 24 hours.
> 2. If a message fails because a mailbox is over quota send a bounce
> and do not ever attept redelivery, delete it immediately.
> 3. Messages that are refused or timeout should never attempt
> redeliver, discard them imediately.
Why should it never happen?
> 4. After 24 hours, expire messages and delete them entirely.
They will be bounced, not deleted!
> 5. Above all, get it out of the queue at all costs!!!
Test the above settings first, and do
exim -qff to enforce a queue-run for ALL mesages.
> In my exim.conf I have this in the RETRY CONFIGURATION section:
> # Domain Error Retries
> # ------ ----- -------
>
> * quota
Thats ok.
> * refused
> * timeout
These both are BAD!
You should not use them, in todays internet you still get
delivery-problems, hotmail comes into my mind or even mail.exim.org was
unreachable for 24h last week.
> * * F,2h,15m; G,8h,2h,1.5; F,24h,8h
>
>
> Exim seems to be completely ignoring me. What is wrong? What can I
> do to correct this? Why won't exim delete expired messages?!?
Set the above settings, the messages that are in your queue for more than
24 hours are most probably frozen messages.
> I'm using exim 3.35-1 on Debian 3.0 (woody). Please please please
> someone help me. As far as I can tell everything I've done is right,
> but exim just doesn't seem to care.
How does your queue-runner is started?
IIRC Debian starts it from crontab by default, which options does it use?
> At this point I'd settle for a nice perl script that will check
> the ammount of time a message has been on the queue and delete
> anything over 24 hours.
First try the above stuff, and please remove the retry-lines for refused
and timeout.
After setting these options in exim.conf run exim -qff and see what
happens.
If you still have much OLD messages in your queue, provide parts of your
mainlog and the relevant parts of a debug-run (exim -d9 -Mc MESSAGEID).
ciao
