I haven't had my morning coffee... so excuse any stupidity.
On Apr 6, 2005, at 7:52 PM, Douglas Otis wrote:
Is there a mailing list discussing this draft?
The null mx record structured in this manner:
* IN MX 0 .
My initial thoughts are that this is a bad idea. Studies by CAIDA and
Paul Vixie have shown that the roots already receive more crap than
legitimate traffic. I'm asking around within our DNS team to see what
they think.
My comments:
1) For most modern mail servers that do not understand the null mx
record, this will result in both an A and AAAA query against the root.
2) There are a lot of pieces of anti-spam software that do not
distinguish between NODATA and NXDOMAIN and there are a lot pieces of
anti-spam software that do. Since this will produce a NODATA which is
the lesser seen case, their will be inconsistency though out the mail
infrastructure where NULL MX is not understood.
3) A very large number of domains already advertise that they do not
send email by using SPF: "v=spf1 -all".
4) A very large number of domains have misconfigured MX records with IP
addresses as the target. Perhaps the null MX is better off redefining
a condition that is currently considered a misconfiguration.
5) Why not "example.org. IN MX 0 _i_send_no_email.example.org."? This
avoids any requery to anything but the local resolvers and example.org.
Personally, I much prefer something like CSV or SPF where there are no
consequences if concept is unknown to receiver.
On Apr 7, 2005, at 5:32 AM, Tony Finch wrote:
When I did testing with . targets on SRV records, it appeared to me
that
BIND was deliberately returning NODATA instead of the negative SRV
record.
Did I interpret its behaviour correctly?
Well, there is no SRV record at the root but the root does exist, so
NODATA seems to be the right answer.
Again... no morning coffee. Everything in this email could be rubbish!
-andy