We have a cluster of Exim machines that are getting a severe kicking.
Our immediate problem is that we can't analyse their spools
as they are so huge, and we are constantly in danger or running out of spool
and losing messages.
I kicked off mailq last night and it hadn't finished after 7 hours <g>
This means that the usual tricks like
exim -Mrm `exim -bpr |grep frozen|awk '{print $3'}`
for clearing mailsicles doesn't work, as the backtick never returns.
Instead I had to resort to:
exim -bpr |grep frozen|awk '{print $3'} >/tmp/icecubes&
and then periodically ran
wc -l /tmp/icecunes
20453
for i in `head -20453 /tmp/icecubes`
do
exim -Mrm $i
done
<wait an hour or so>
wc -l /tmp/icecubes
48023
for i in `tail -28000 /tmp/icecubes`
do
exim -Mrm $i
done
Yuck (but helps a little)
Our configure files contain:
ignore_bounce_errors_after = 12h
timeout_frozen_after = 1d
but we have frozen messages that are days old, it appears that
the queue runners don't have time to expire the frozen messages?
This is despite:
queue_run_max = 60
So, to sum up, we're getting a beating, and can't see the spam for the frozen
bounces.
I wondered if there is a way to stop exim bothering to freeze messages at all,
can it just delete the bounce there and then?
I can't see any other way to stop our spools overflowing....
'Buy more disk' might help short term, but to be honest I doubt it would
for long.
Thanks!
--
Serving coffee on aircraft causes turbulence.
Rasputin :: Jack of All Trades - Master of Nuns
