Re: [exim] detection of "<>" in case of spam.

On Mon, May 13, 2013 at 11:25:35PM +0100, David Woodhouse wrote:
> On Mon, 2013-05-13 at 22:37 +0400, Evgeniy Berdnikov wrote:
> > Pointless. Considerable amount of junk traffic consists of real
> > and valid bounces, generated to spam mails with forged addresses.
> > If you have several old domains with old addresses, some of them
> > might be used by spammers, and you would get bounces to these
> > addresses when other sites bounce spam mails.
>
> If they are indeed "old" addresses, surely you never send email from
> those addresses any more? And thus you never need to accept bounces *to*
> this addresses?

In a company I'm working for I hold exactly the same policy as you wrote:
users are encouraged to use old addresses only for incoming mails.
Bounces to old addresses are not accepted. It really works.

> In fact, you don't really need to accept bounces to the *new* addresses
> either. I don't accept bounces to dwmw2@???, for example,
> because I never send MAIL FROM:<dwmw2@???>.

From: in mail headers and env-from are different? Looks interesting.

But it seems too complex for simple user mind, and probably could not be
configured for bare Outlook or FireFox without envelope rewriting on the
server side... Have you ever run such a service in production?
P.S. I do understand that VERP/BATW/etc are typical for mail lists.
--
Eugene Berdnikov