Re: [Exim] Performance bottleneck scanning large spools.

On Wed, Jan 19, 2000 at 03:18:40PM -0000 John P Connor wrote :

> When I had a HUGE queue (86,000) last week, I managed to overcome it (or at
> least I think I did) by queueing everything with queue_only & running
> exim -q once a minute, together with 5 exim -R 200 to force through mail I'd
> re-written to a local host (with 200 in its name). This seemed to work
> pretty well. Running exiwhat now and again told me there were quite a few
> exims which were "running queue: waiting for <message id>".

;-) I think we have all done similar things! I normally avoid exim -q or exim
-qf when the queue has gotten to big for some reason. In fact I will
sometimes stop the queue runner and just do lots of exim -R whatever to clear
it all down.

You seem to get a situation where one awkward message can block up the queue
runner and the next one sits behind it waiting for it to finish and the next
one ..... Obviously after a while the retry time gets set and the message
stops holding things up but, the other active queue runners don't seem to
notice (hence stopping the queue).

The other thing I watch out for is, if you do an exiwhat, is deliveries that
have just hung there for a very long time (hours or even days). These can
cause problems as well and I normally kill them off.

Truth is though, mostly, you are just as well to leave nature to take its
cause, as long as the systems are ticking over and your net feed is working,
it will all go in the end.

--
Ian