Re: [Exim] How to deal with 10 fold increase in mail traffic

Hi Sujit, on Fri, 13 Feb 2004 14:34:44 -0000 you wrote:

> The problem is our MTA is a gateway. It handles the traffic for
> m25lib.ac.uk, psi.org.uk etc. It does not have local users. Depending
> on various rewrite rules, it just passes the mail to some other (less
> powerful) host.

Then ideally, if possible, you want to do SMTP callouts to allow the
frontline MX to verify the recipients. (something like verify =
recipient/callout=25s). If the backend machines run stupid software like
Exchange, you might need to do an LDAP lookup or something.

You really, really want to do this anyway, because otherwise if a spammer
sends mail from a forged address to one of your domains, you'll end up
creating collateral spam by bouncing back to the faked sender. (How many
times a day does this come up I wonder :) . In other words, you'll end up
inflicting the same as that which you're experiencing now (i.e. getting
lots of bounced junk you didn't send) on others.


So the solution both for your own benefit and everyone else's (being a
responsible admin), is to work out a scheme whereby your frontline MX
*does* know which are valid recipients and which aren't. This really is
essential. If the schemes above (SMTP callout, LDAP) don't work in your
situation, you may need to fall back onto something more generic (e.g.
generating a flat file user list at intervals from whatever data source
defines the valid recipients for the back-line servers).


Tim