On Jun 8, 2014, at 5:31 PM, Terry Zink <[email protected]> wrote:
>> Hector Santos <[email protected]> wrote: >> >>> It is mentioned in Section 6, but the mention there doesn't even say >>> that it's the DMARC result that's supposed to be recorded. That bit >>> at least needs to be fixed. >>> >>> Anyone else have a comment? > >> Only that it goes back to the similar SPF thing regarding dynamic >> rejections. So to be consistent for DMARC: >> >> DMARC POLICY A-R Trace Guideline >> >> REJECT --> N/A see 55x reply codes. >> QUARANTINE --> SHOULD record with A-R. >> NONE --> SHOULD record with A-R. > > In our implementation, we also plan to indicate the following scenarios: > > 1. DMARC fails but the pct value indicated not to take action (i.e., pct=80, > this message failed but is in the 20% of cases not to take action) at which > point we would stamp "dmarc=fail action=pct.quarantine" > Sounds reasonable. Haven't implemented DMARC yet, but I got the fast read understanding "pct" was for reporting triggers. It is not factored in when a reject is triggered? > 2. DMARC fails and the action is reject but we overrode the requested action > (i.e., for a mailing list) at which point we would stamp "dmarc=fail > action=o.reject" Is "action" is a registered tag? > Others may feel that this is not necessary (of think this syntax is not clear > enough) but we do it for troubleshooting -- "This domain's DMARC action is > reject. Why is it in my junk folder?" For ADSP, the same level of diagnostic consideration was done because we were still in recording mode, not yet enabling honoring the discard policy yet. DMARC changes all this now. FedEx.com and other high value domains use both ADSP discard and now DMARC rejects, so when DMARC is added, both will have a common handling engine. -- Hector Santos http://www.santronics.com _______________________________________________ dmarc mailing list [email protected] https://www.ietf.org/mailman/listinfo/dmarc
