On Fri, May 04, 2001 at 02:57:20PM +0100, EXIMUsers@??? quothed:
> As you may have seen from my "Exim performance on large mail systems"
> posting a couple of days ago we've just switched over to using 'queue_only',
> 'queue_run_max = 200' and starting exim with '-bd -q1m' as a number of other
> people have suggested to combat high loads...
That's starting an awfull lot of queue runners. My guess is that you are
not starting one a second, but more like one every 15/20 mins (from
times below) because of your 'queue_run_max = 200' setting.
> This is causing major problems though, now mails are taking up to 20 minutes
> to even be reached by a queue runner - I can only guess this is because we
> have a lot of mails in our queue and the randomized queue running process is
> just working against us and some mails aren't getting looked at for a long
> time?? There are no connectivity issues to this remote site, i'm simple
> sending a mail to a <nonexistant-localpart>@remotedomain.com to see how long
> it'll take to bounce and the mail is sitting in my outbound mail queue for
> 15 to 20 minutes before a delivery is even attempted.
How big is your spool? Is it tens, hundreds or thousands of messages?
remember that exim needs to open *all* the H files first to see what it
does with a mail. This takes time on large queue.
> Is there anyway I can speed this process up without reverting back to
> attempting deliveries as soon as mails are received (which results in >800
> exim processes loading).
But 'queue_run_max = 200' will limit the number of processes anyway.
How many mail servers do you have? Adding a couple may well help you.
--
www.kierun.org
Yann@??? Use Pretty Good Privacy.
