Liam:
> On Sun, Nov 7, 2010 at 10:01 AM, Wietse Venema <[email protected]> wrote:
>
> > It sounds like you will be sending all mail to the same outbound
> > proxy, meaning that Postfix has no clue what destinations are up
> > and what are down, because that is hidden by your outbound proxy
This holds for all delivery mechanisms that need to be aware
of the final destination.
> I can support per-node domains, e.g. node-id.mailhost.xyz
I suppose some wild-card DNS record will do the trick.
> > The number of messages per client is not the problem. The problem
> > is having 15% of the daily email volume sitting in the mail queue.
> > These deliveries are relatively expensive and can end up dominating
> > the over-all system load.
> >
>
> Relatively expensive even if they're prompted with ETRN for a single node's
> domain?
The optimized fast path is:
incoming queue -> active queue -> remove
The non-optimized slow path is:
incoming queue -> active queue -> deferred queue
[try to deliver while the destination is down, with exponential back-off]
deferred queue -> active queue -> deferred queue + append fast-flush file
deferred queue -> active queue -> deferred queue + append fast-flush file
...
[destination comes up, proxy sends ETRN]
read fast-flush file
deferred queue -> incoming queue
[ETRN completed]
incoming queue -> active queue -> remove
This is why 15% slow mail can cost more than the 85% that is
delivered immediately.
And that is with Postfix's fast-flush optimization. It will be
worse with MTAs that have to examine every queue file.
Wietse
