On 05/27/2013 05:15 PM, Matt Simerson wrote:

Are you intending to inform all implementers and domain owners of this limitation with admonitions to search the list archives?

Searching is not required; this has been thrashed out so often and such length that any 10 randomly selected messages from the archive will likely contain several that argue one angle or another of this issue.

Your points about the weakness or absence of this information at several key places off-list are sensible.

Sorry, I missed the URL to the "Build an appropriate DMARC record" form that has the 
distilled wisdom of someone that has read every post to this list. Especially the official DMARC 
record creator tool with the "Do you have users that subscribe to email lists?" radio 
button that disables the rest of the form when I choose Yes.

May I suggest, as delicately as possible, that you [re-]familiarise with the obligations of all posters to this list and, in particular, with the requirement to remain cordial? http://www.dmarc.org/note_well.html

When I sent emails from my DMARC enabled tnpi.net domain to this email list, I 
got back reports from google and yahoo that they're blocking them.

513            google.com              tnpi.net  2013-05-22 17:00:00
<snip>
   | --  30                    2001:1890:123a::1:1e        reject    fail    
fail                        mail.ietf.org  forwarded  mailing_list
   | --   1                    2607:f0d0:2101:bb::2        reject    fail    
fail                   ericindustries.net  forwarded  mailing_list
   | --   1                           113.56.245.52          reject    fail    
fail                113.56.arpa.hb.cnc.cn
   | --   1                          209.85.217.176        reject    fail    
fail             mail-lb0-f176.google.com  forwarded  mailing_list
   | --   1                          208.75.177.101          none    pass    
pass                  mail.theartfarm.com
<snip>
   | --   1                 2607:f8b0:4001:c03::234        reject    fail    
fail             mail-ie0-x234.google.com  forwarded  mailing_list
   | -- 113                           208.69.40.157        reject    fail    
fail                  medusa.blackops.org  forwarded  mailing_list

The only rejection listed above is the one by cnc.cn. The rest of the "assessed policy is reject" cases are situations in which the Receiver has locally over-ridden the Domain Owner's DMARC Policy, in each case because a reliable forwarder (specifically a mailing list) has been recognised.

It's worth noting that my observation that it's reasonable for a Receiver to map reliable forwarders and perform this kind of override does _*not*_ mean that it is reasonable for Domain Owners to assume that Receivers will do so. All participants in DMARC (and in email generally) need to assume that a material fraction of their communication partners will do unwise things.

It seems apparent that google detected that medusa is a list server, but they 
still applied my reject policy. When the receiver whitelists a list server, the 
results look like this:

519               163.com              tnpi.net  2013-05-23 09:00:00
   | --   1                           208.69.40.157          none    fail    
fail  mailing_list( forwarded by a white-listed mailing list )  
medusa.blackops.org

That's how 163.com indicates it, sure. Google's indication of this outcome is closer to the spec and is you've shown above.

Had there been a warning label somewhere that DMARC + mailing list = BROKEN 
EMAIL, I'd have subscribed to this list with a different address.

As above, this is a good point and needs addressing. There's an even bigger warning label needed though: if your domain is not being heavily spoofed, you need to not use p=reject/quarantine at all. The reason is that the moment that you do so, a range of complications and dilemmas arises that require extensive domain knowledge to deal with. If you don't have the problem (being heavily spoofed), then you are better off avoiding the collateral damage entirely.

Now I'm exploring ways to work around the problem, so that I can still prevent 
phish from being sent from my domain.

How much phish is being sent from your domain at present?

Roland suggested not signing the Subject and body with DKIM, which is 
potentially another.

(cough) That was meant to be a straw man in that it was supposed to expose an obvious flaw: if you send a _*single*_ message signed this way to a publicly archived/subscribable list then a phisher can send an unbounded number of fully authenticated messages containing whatever he wishes until you actually delete mar2013._domainkey.tnpi.net from the DNS.

Sorry about that.

I'm also curious if the DMARC incurred deliverability failures will cause list 
users to inadvertently get unsubscribed. That would be an interesting side 
effect.

Enter the next dilemma for Receivers: silent discard causes message loss. SMTP refusal to an unmapped-but-reliable forwarder is likely to cause list unsubscription.

There is the potential for collateral damage almost everywhere.

- Roland

--
  Roland Turner | Director, Labs
  TrustSphere Pte Ltd | 3 Phillip Street #13-03, Singapore 048693
  Mobile: +65 96700022 | Skype: roland.turner
  [email protected] | http://www.trustsphere.com/

_______________________________________________
dmarc-discuss mailing list
[email protected]
http://www.dmarc.org/mailman/listinfo/dmarc-discuss

NOTE: Participating in this list means you agree to the DMARC Note Well terms 
(http://www.dmarc.org/note_well.html)

Reply via email to