On 25 Jan 2008, at 13:17, Peter Kirk wrote:
> That's why I was hoping if I can avoide using a queue for messages and
> just send them out straight away, then this would send them faster as
> they would not sit in a queue for an hour or so but I just wanted to
> know if this would kill my machine?
Ah - so my reading of this thread gives the following implications:-
1. The XP box fires a whole pile of messages, each with
one recipient (assumption made on the basis that you
say your queue gets to 15K plus), probably sending as
many messages in an SMTP session as it can.
2. The majority of these messages end up sitting in the queue
for an hour (ish) before anything happens.
I would guess that your queue run is approximately hourly, or at
least not very frequently.
Exim always puts mail into the queue - by putting it on stable
storage (ie to disk with a sync op) it can guarantee (as far as
it can) that the message won't get lost due to system outage.
Unless your box is suffering from severe overload (and with
VMWare ESX you have some decent tools to check on the loading)
I'd suggest you want to run queue runners more frequently.
You could increase smtp_accept_queue_per_connection but that
risks overloading the box if it receives several thousand
mails at once.
Running a batch more queue runners - especially around the time
that the mass mailing is happening, will probably help a lot.
But check the box load to ensure its not being swamped.
Nigel.
>
--
[ Nigel Metheringham Nigel.Metheringham@??? ]
[ - Comments in this message are my own and not ITO opinion/policy - ]
