Hi Andrew,

While preparing our incident report, we reviewed all previous 
communications and noticed a small typo in our previous message. The SC46 
action item was flagged on 2021-04-29, rather than 2021-04-09.

We apologize for any confusion this may have caused.
Brett L
Google Trust Services

On Friday, September 3, 2021 at 6:34:47 PM UTC-4 Brett L wrote:

> Hi Andrew,
>
> As mentioned earlier the CPS update including this change was not 
> published earlier because the update was bundled together with the annual 
> CP/CPS update. As you know, GTS also had multiple incidents open around the 
> time of these changes, so we were engaging a number of additional partner 
> teams and stakeholders for reviews with the goal of ensuring we covered all 
> updates and improvements in one pass. Wider participation in the update 
> helped highlight additional areas to simplify and clarify language, but 
> this also took additional time to complete. 
>
> We flagged SC46 as an action item on 2021-04-09 and the change to our CPS 
> was made in draft form a few days later (2021-05-13). Though the automation 
> in question was in active development at the time, it was not completed. 
> When SC46 was published a few weeks later (2021-06-02), the review process 
> for requirement changes identified it as having been addressed because at 
> the time both the associated code changes had been deployed and the CPS was 
> in the process of being approved for release. 
>
> Though we always try to land documentation changes with corresponding code 
> changes sometimes it is simply not always feasible.  That being said upon 
> review we feel the 30 day delay that was introduced as a result of the 
> bundling of other changes together is too long and have now filed an 
> incident report 1729097 to capture this event.
>
> As for https://bugzilla.mozilla.org/show_bug.cgi?id=1706967 it was 
> different in a few ways. The first is that Mozilla Root Store Policy under 
> section 2.2 (3) has a clear expectation that "The CA's CP/CPS must clearly 
> specify the procedure(s) that the CA employs, and each documented procedure 
> should state which subsection of 3.2.2.4 it is complying with." In that 
> particular case, the CPS included a forbidden validation method. 
> Additionally we had missed that the change in numbering had not been 
> reflected in our CPS. In this case, the changes to remove the operator 
> exception were complete, the documentation was in progress and our tracking 
> and automation tooling covered both. The second difference is that 
> 4.9.1.1.7 obligated us to revoke the certificate because although our 
> practices were conformant with current regulations, the CPS no longer had a 
> reference to an allowed domain control verification scheme.
>
> Brett L
> Google Trust Services
>
>
> On Thursday, September 2, 2021 at 11:07:27 AM UTC-4 Andrew Ayer wrote:
>
>> On Wed, 1 Sep 2021 14:21:47 -0700 (PDT) 
>> Brett L <[email protected]> wrote: 
>>
>> > Hi Andrew, 
>> > 
>> > Thank you for the questions and checking on details. 
>> > 
>> > We removed the option to use the DNS operator exception from our 
>> > secondary CA platform on 2021-05-13 (60 days before the ballot 
>> > changes went into effect, see timeline below). Our primary CA 
>> > platform has never used it. 
>> > 
>> > In May, we conducted our annual CPS review and prepared several 
>> > updates including the one that removed the exception from Section 
>> > 4.2.4. It was not published earlier because the update was bundled 
>> > together with other changes in one revision. 
>> > 
>> > We did not file an incident because removal of the DNS operator 
>> > exception was identified and acted upon well ahead of the deadline. 
>> > The CPS update was also started before SC46 became effective. We 
>> > regret it was not published prior to the effective date. 
>>
>> The inaccurate CPS is a compliance violation, even if the DNS 
>> operator exception was not in use. 
>>
>> This is the same basic scenario as 
>> <https://bugzilla.mozilla.org/show_bug.cgi?id=1706967> - in that case, 
>> GTS' domain validation practices were compliant with the BRs, but GTS' 
>> CPS did not accurately reflect GTS' practices. 
>>
>> It's troubling to see a recurrence of that scenario, and your email 
>> doesn't provide much of an explanation how this happened. 
>>
>> 1. Why was the CPS change not published in May given that 
>> <https://bugzilla.mozilla.org/show_bug.cgi?id=1706967> made clear the 
>> importance of having an accurate CPS? 
>>
>> 2. When SC46 was merged into the BRs, did the automation described in 
>> <https://bugzilla.mozilla.org/show_bug.cgi?id=1706967#c11> create an 
>> internal ticket for it? If not, why not? If it did, how did GTS respond 
>> to the ticket and why did the response not detect the CPS 
>> non-conformance? 
>>
>> Regards, 
>> Andrew 
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"[email protected]" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/a/mozilla.org/d/msgid/dev-security-policy/775ab731-6ca7-493b-94c6-660d0f825106n%40mozilla.org.

Reply via email to