Andreas Grip <[EMAIL PROTECTED]> wrote:
> >
> > I don't think this is a great idea; it means you have to accept every message,
> > then scan them, then generate late bounces, instead of rejecting them during
> > the initial SMTP conversation.
>
> qmail-scanner do not reject them, it just bounce them.
I think you're mistaken, although I don't use qmail-scanner. Issuing a 4xx or
5xx code after DATA _is_ rejecting a message -- it's also a bounce, although
if it's done during the SMTP conversation, the sending MTA is responsible for
generating the bounce message.
> And what diffrent should that make if the bunce is a few minutes late? It
> will be late for the sender anyway because they use their ISP:s smtp server
> and the mail will be sended from that to my smtp server that scan the mail.
There's a big difference. See above. Late bounces have to be generated by
your MTA and delivered; if the message is bounced during the initial SMTP
conversion, the bounce message is the responsibility of the sending MTA, not
the receiving one.
> > What problem are you trying to solve? Why do you think making the SMTP
> > client wait a minute or two is a bad idea?
>
> Well, a smtp-server receiving a lot of mail can reach the limit of maximum
> allowed simultanius connection. If the smtp server close the connection
> faster there will be more time over and the server is able to receive more
> mail. So I think a server, that are faster with closing the connection
> should be more efficient.
Profile, don't speculate. You're trying to solve a problem that doesn't
exist.
Charles
--
-----------------------------------------------------------------------
Charles Cazabon <[EMAIL PROTECTED]>
GPL'ed software available at: http://www.qcc.sk.ca/~charlesc/software/
-----------------------------------------------------------------------