R: Q. about spam directed towards highest MX Record?

2006-10-18 Thread Giampaolo Tomassoni
> | Just to clarify here You are talking about doing something like:
> |
> | domain.com   1200   IN   MX   10  smtp-1.domain.com
> | domain.com   1200   IN   MX50  smtp-2.domain.com
> |
> | You all are saying that most of the spam should be coming in MX 
> 50 right?
> |
> | I have to admit I've tried this, but it seems like mail continues to
> | come into the MX 50 even when the primary servers are available.Is
> | it not correct that the 50 should NOT be tried until the 10 is
> | unavailable?  Or do I have that backwards?
> 
> You have it right.  Unfortunately, mail still hits the lowest 
> priority server based on my experience 
> even when the Primary is up and running.

Are you using something like greylisting or maybe your primary is often under 
heavy load?

giampaolo



R: Q. about spam directed towards highest MX Record?

2006-10-18 Thread Giampaolo Tomassoni
> > We tried that and had problems with some clients (the business client
> > not the mail client). Seems a lot of Exchange servers will try the
> > lowest priority MX for some reason, and then never try the highest, just
> > fail.
> >
> > With the current setup a valid message will eventually get through.
> >
> > DAve
> 
> Isn't that how it is suppose to work?  Try the lowest first?

Yes, it is:

MX 10 primary-mx.domain.tld

has generally to be checked before falling back to:

MX 20 secondary-mx.domain.tld

Lowest number is highest priority.

giampaolo



R: Q. about spam directed towards highest MX Record?

2006-10-18 Thread Giampaolo Tomassoni
> > > You have it right.  Unfortunately, mail still hits the 
> > lowest priority 
> > > server based on my experience even when the Primary is up 
> > and running.
> 
> Or, even better, point it at an unused IP on your network.
> (don't point it at 127.0.0.1, that will get you blacklisted in the
> rfc-ignorant invalid mx list)
> 
> That way, no bandwidth used except for a tcp syn every now and again.

... and spammers would have to wait for a tcp timeout before giving up there. :)

giampaolo