[Servercert-wg] [EXTERNAL] Re: Voting Begins for Ballot SC46: Sunset the CAA exception for DNS Operator

Wed May 26 21:40:29 UTC 2021

There’s nothing that prohibits a correctly written ballot (new ballot number, same endorsers) from running concurrently with the failure of the current one.  So the practical effect is only a delay of a week.  The longer it takes us to decide this is worth doing correctly, the longer the delay will be.

-Tim

From: Ryan Sleevi <sleevi at google.com> 
Sent: Wednesday, May 26, 2021 4:32 PM
To: Tim Hollebeek <tim.hollebeek at digicert.com>
Cc: Bruce Morton <bruce.morton at entrust.com>; CA/B Forum Server Certificate WG Public Discussion List <servercert-wg at cabforum.org>
Subject: Re: [Servercert-wg] [EXTERNAL] Re: Voting Begins for Ballot SC46: Sunset the CAA exception for DNS Operator

Sorry, I realized we never addressed with "withdraw a ballot early" within the Bylaws.

This means the possible outcomes are:

*	Vote against this ballot (one week), start a new discussion period (one week), start a new voting period (one week). So the net delay is 3 weeks, not one week.
*	Accept this ballot as is (one week for voting), and if it is adopted, bundle any cleanups as part of a belated Spring cleanup ballot, which I've already been working on items to tackle

My clear preference is for the second option, because I think where I disagree with Tim is that the removal of this exception sooner, then later, is a net positive for security. While based on our understanding, the incidents with this CAA exception have already lead to CAs relying upon it to fully shift to checking CAA all the time, that's still something to make progress on. The only CAs that would run any risk with the language would be CAs still relying on the CAA exception, which, to the best of my knowledge, are none, but unfortunately, we can't actually rely on that until this ballot is adopted.

On Wed, May 26, 2021 at 4:25 PM Tim Hollebeek <tim.hollebeek at digicert.com <mailto:tim.hollebeek at digicert.com> > wrote:

I agree that the practical effect, if this ballot were to pass, is that the effective date would end up being when the guidelines get published.  The effective date will be functionally inoperable, as it will be in the past at the time it becomes official.  I am a bit concerned about the confusion that may cause, and I think at a minimum we should all agree that this is a situation we should try to avoid in the future.

I personally think it would be best to remove the effective date and restart the discussion period, just to avoid unnecessary ambiguity and misinterpretations.  I don’t think there’s any need to be hasty here, and I would prefer that we provide clear and non-contradictory information about when the effective date is.

-Tim

From: Servercert-wg <servercert-wg-bounces at cabforum.org <mailto:servercert-wg-bounces at cabforum.org> > On Behalf Of Bruce Morton via Servercert-wg
Sent: Wednesday, May 26, 2021 3:57 PM
To: Ryan Sleevi <sleevi at google.com <mailto:sleevi at google.com> >
Cc: CA/B Forum Server Certificate WG Public Discussion List <servercert-wg at cabforum.org <mailto:servercert-wg at cabforum.org> >
Subject: Re: [Servercert-wg] [EXTERNAL] Re: Voting Begins for Ballot SC46: Sunset the CAA exception for DNS Operator

I don’t feel strongly on this item.

Bruce.

From: Ryan Sleevi <sleevi at google.com <mailto:sleevi at google.com> > 
Sent: Wednesday, May 26, 2021 3:49 PM
To: Bruce Morton <Bruce.Morton at entrust.com <mailto:Bruce.Morton at entrust.com> >
Cc: CA/B Forum Server Certificate WG Public Discussion List <servercert-wg at cabforum.org <mailto:servercert-wg at cabforum.org> >
Subject: Re: [EXTERNAL] Re: [Servercert-wg] Voting Begins for Ballot SC46: Sunset the CAA exception for DNS Operator

An excellent question, and an unfortunate situation with respect to me being out of office and missing the start of the voting period.

The short answer is "no", because the BRs don't apply retroactively to past events, only to new events. This was similar to the discussion had about validation methods and reuse of data.

Basically, the effect of an effective date in the past is that it behaves "as if" it's not effective until the BRs are adopted, and on a go-forward basis. For something like the CAA exception, what this means is that if a certificate was issued on 2021-07-01, and this ballot wasn't effective until 2021-07-02, that certificate would have complied with the then-current version of the BRs, and not have been misissued. That's because, at least with respect to this ballot, it's touching on an event that happens "in the present"; it can certainly get messier with other ballots.

Equally, this also means that if you issued a pre-certificate on 2021-07-01, and relied on the CAA exception because DNS operator, and then issued the actual certificate on 2021-07-03, you'd also be fine: because the certificate issued on 2021-07-03 was relying on the pre-certificate exception, not the DNS operator exception, when it was issued, and thus complies with the effective version of the BRs on the date the certificate was issued.

The alternative would be to remove the "Effective 2021-07-01" language, and restart the discussion period. The net effect would be that it'd shift the effective date out a week. Sorry, I should have highlighted this rationale for why I didn't, because it's understandable to bring it up. However, because we've consistently said "The version of the BRs that apply are the version of the BRs when the certificate is issued", it wouldn't be misissuance.

That said, this is also why the discussion in validation WG is trying to find ways to make sure that "When the certificate is issued" is accurately reflected in the certificate's notBefore, since otherwise, complexity arises for relying parties and CAs.

If you feel strongly, though, I can explore restarting the ballot. It was entirely my fault for not setting up a scheduled send. There was some brief discussion of this problem two calls ago, during the ballot update, to draw attention to the possibility.

On Wed, May 26, 2021 at 3:02 PM Bruce Morton <Bruce.Morton at entrust.com <mailto:Bruce.Morton at entrust.com> > wrote:

Quick question about the voting period. Is there an issue if the ballot is effective prior to the IPR period concluding?

Thanks, Bruce.

From: Servercert-wg <servercert-wg-bounces at cabforum.org <mailto:servercert-wg-bounces at cabforum.org> > On Behalf Of Ryan Sleevi via Servercert-wg
Sent: Wednesday, May 26, 2021 2:30 PM
To: Ryan Sleevi <sleevi at google.com <mailto:sleevi at google.com> >; CA/B Forum Server Certificate WG Public Discussion List <servercert-wg at cabforum.org <mailto:servercert-wg at cabforum.org> >
Subject: [EXTERNAL] Re: [Servercert-wg] Voting Begins for Ballot SC46: Sunset the CAA exception for DNS Operator

WARNING: This email originated outside of Entrust.
DO NOT CLICK links or attachments unless you trust the sender and know the content is safe.

  _____  

Unfortunately, I realized belatedly that I forgot to clearly indicate the Voting End Time.

As such, the previous mail did not officially start voting. Thankfully, as no votes were received, I think we can just say I didn't start it correctly?

Please find the corrected announcement below:

This email begins the voting period for Ballot SC46: Sunset the CAA exception for DNS operator

Purpose of Ballot:

This Ballot addresses security issues with Section 3.2.2.8 regarding CAA checking.

Currently, Section 3.2.2.8 permits a CA to bypass CAA checking if the CA or an Affiliate of the CA is the DNS Operator. This term is referred to through RFC 7719, and involves a precise technical definition regarding how a zone's authoritative servers are configured and expressed (e.g. NS records). While this allows a CA to skip looking up the CAA record, it does not absolve them of the need to look up these other records on every issuance.

As practiced by CAs, this has clearly caused some confusion. For example, some CAs have incorrectly implemented policies that determine they're authoritative based on self-assertion that they are authoritative, which is not consistent with the current requirements.

To avoid these issues, this sunsets the CAA exception on 2021-07-01 for the DNS Operator, simplifying the requirements and reducing ambiguities for CAs performing validation.

The following motion has been proposed by Ryan Sleevi of Google and endorsed by Ben Wilson of Mozilla and Jacob Hoffman-Andrews of ISRG/Let's Encrypt.

It can be viewed on GitHub as https://github.com/cabforum/servercert/pull/271 <https://urldefense.com/v3/__https:/github.com/cabforum/servercert/pull/271__;!!FJ-Y8qCqXTj2!NFeHHMg2M0PSERtj03rqrCoxas3jZqEeftaCsg3iAoFNIJ7Gmq5rzCN_3XxzPiKIUOI$> 

-- MOTION BEGINS --

This ballot modifies the “Baseline Requirements for the Issuance and Management of Publicly-Trusted Certificates” (“Baseline Requirements”), based on Version 1.7.4:

MODIFY the Baseline Requirements as specified in the following Redline:

https://github.com/cabforum/servercert/compare/47248d77d371356780b08cfa971b26d88d704ca8..6d34b1d51f645912d2237d5d4b46f4a49e8352ed <https://urldefense.com/v3/__https:/github.com/cabforum/servercert/compare/47248d77d371356780b08cfa971b26d88d704ca8..6d34b1d51f645912d2237d5d4b46f4a49e8352ed__;!!FJ-Y8qCqXTj2!NFeHHMg2M0PSERtj03rqrCoxas3jZqEeftaCsg3iAoFNIJ7Gmq5rzCN_3XxztHcKH2U$> 

-- MOTION ENDS --

This ballot proposes a Final Maintenance Guideline.

The procedure for approval of this ballot is as follows:

Discussion (7+ days)

Start Time: 2021-05-13 20:00:00 UTC
End Time: 2021-05-26 14:00:00 UTC

Vote for approval (7 days)

Start Time: 2021-05-26 18:30:00 UTC
End Time: 2021-06-02 18:30:00 UTC

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cabforum.org/pipermail/servercert-wg/attachments/20210526/3c9ca0aa/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4940 bytes
Desc: not available
URL: <http://lists.cabforum.org/pipermail/servercert-wg/attachments/20210526/3c9ca0aa/attachment-0001.p7s>