[Gllug] Rejecting mail at backup MX

Alistair Mann alistair at lgeezer.net
Wed Feb 11 14:28:56 UTC 2004


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Thus spaketh Bruce Richardson on Wednesday 11 February 2004 1:00 pm:
> On Tue, Feb 10, 2004 at 08:36:15PM +0000, Alistair wrote:
> > > Bollocks.  Show me the RFC that says this.  I'd be impressed if you
> > > could, because the concept of a backup MX doesn't even exist in the
> > > RFCs and this idea that "backups" shouldn't do X or Y is entirely
> > > arbitrary and made up.
> >
> > Very true, entirely arbitrary and made up for the simple reason that It
> > Works.
>
> The fact that you find it convenient gives you absolutely no grounds at
> all to say to somebody else "I think you have understood what a backup
> is for" or "You shouldn't be rejecting mail from your backup".  It's
> tosh.

In hindsight, I ought to have said "I think I have misunderstood what you 
think a backup is for", it's true.

> > > Why impose this arbitrary rule that does no good
> > > but helps spammers and virus writers?
> >
> > Exactly how does it help spammers and virus writers to have thier spew
> > deleted on a primary instead of a backup? It doesn't.
>
> It bloody well does.  A mail system that is receiving mail directly from
> a virus or spammer can impose checks that can't be imposed by a box
> receiving mail from a relay.  You can detect and thus reject a higher
> amount of illegitimate mail if all your mail exchangers are imposing
> policy on incoming mail.  More importantly, you almost entirely
> eliminate the problem of spurious NDRs, because you aren't accepting
> illegitimate mail for delivery at all.
>
> If, on the other hand, all but one of your mail exchanges is accepting
> all traffic to your domains and leaving it to other systems to accept or
> reject, not only are you wasting a lot of your own bandwidth but you
> then *have* to spam a lot of innocent people with spurious NDRs, because
> a) every message that you accept for delivery and then decline to
> deliver is required to generate an NDR and b) you now have no way of
> telling the spurious NDRs from the real ones.

Good point, a situation that for me hasn't thus far come up.

> > Doing so considerably eases administration for me, though.
>
> The fact that you don't want to do something doesn't make it a bad idea.

Nor did I say it was a bad idea, though you certainly seemed to have taken it 
that way.

> I put quite a bit of effort into ensuring that my mail systems rejecting
> invalid mail while making sure that a) legitimate senders are informed of
> rejections and b) we aren't spamming people with spurious NDRs about
> spam and viruses that they didn't send.  It's only good citizenship,
> given the amount of hassle that Sobig and MyDoom and spammers are
> causing, not to add needlessly to the noise.  If you are happy to spam
> people with useless NDRs there's nothing I can do to stop you but it's
> not a position I can respect, particularly when you then start lecturing
> people who are making and effort and telling them that they are wrong.

Well, I'm sorry you got the idea I thought you were doing wrong: it occured to 
me that you were expending a huge amount of effort dealing with a problem on 
the backup that has -- in all situations I've encountered -- been 
satisfactorily dealt with on the primary, and observed that fact. 
- -- 
Alistair
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)
Comment: As seen at http://search.keyserver.net

iD8DBQFAKjwrEz+/jt85AfsRAhOjAJoDzty1C09XJFw03pucYRSbCZtnpgCfUYab
SLjJMMGSWkaP/lLpOEa3yA8=
=XX8g
-----END PGP SIGNATURE-----

-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug




More information about the GLLUG mailing list