My 2 cents...

The presence or absence of a PTR record is, to me, like a reverse-DNS Literacy 
Test.  

History records that Literacy Tests didn't fare too well, as voting 
requirements, in the "real" (non-IT) world. In fact, they were just thin 
pretexts for racial bigotry, recognized as such, and eradicated. 
http://en.wikipedia.org/wiki/Literacy_test#voting

Do we think that a reverse-DNS Literacy Test will fare any better, as a way to 
ensure that only "fine, upstanding, properly-complected" TCP/IP-capable devices 
send mail? I'm all for combatting spam, and I'm all for populating reverse DNS, 
where it makes sense for certain device classes or types, but I just don't see 
the value of linking those things together. It's a bad marriage.

Lastly, what causes us, as a private enterprise, more heartburn in practice, is 
not missing PTRs, but *mismatched* PTRs (the RDATA of the PTR resolves to a 
*different* A/AAAA record). The compulsive drive to machine-generate PTRs, 
combined with sloppy maintenance processes and controls, actually exacerbates 
*that* problem. It would actually be better "if you can't do it right, then 
don't do it at all", with respect to PTR-record creation, among many of our 
hosting providers.

                                                                                
- Kevin

-----Original Message-----
From: DNSOP [mailto:[email protected]] On Behalf Of Paul Vixie
Sent: Thursday, May 14, 2015 4:31 AM
To: Shane Kerr
Cc: [email protected]
Subject: Re: [DNSOP] Rejecting Practice for Theory (was Re: relax the 
requirement for PTR records?)



Shane Kerr wrote:
> ...
>
> However, as far as I can tell everyone insisting that PTR is important 
> is arguing that the world would be a better place if every endpoint on 
> the Internet was equal.

if by "equal" you mean "so expensive that it won't be an open relay, won't get 
infected with relaying malware, will be monitored, will be upgraded, and its 
owner will accept complaints about it" then yes i'd argue that the world would 
be a better place if every endpoint on the internet was "equal".

however, it ain't so, and ain't ever gonna be so. most devices are cheap, 
mobile, proprietary shankware. IoT is going to accelerate that trend 
unimaginably (unless you have an exquisitely dark imagination.)

so, given that endpoints aren't equal, and that by sheer mass of numbers, most 
endpoints are dangerous to themselves and others, i'd like some method by which 
i, as an SMTP responder, can tell the difference.

lack of PTR, and machine-generated PTR, are pretty good telltales, just 
expensive in the case of machine-generated. any rules change should either make 
that situation better, or at minimum, not make it worse.

but i think i'm offering a minor summary-correction, in that i'm not arguing 
for endpoint equality. rather, i'm arguing that in the proved absence of such 
equality, we have other steps we MUST take as receivers.
("be liberal in what you accept" stopped being a good idea in 1995 or so when 
commercialization/privatization took hold.)

--
Paul Vixie

_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to