[Acme] Secdir last call review of draft-ietf-acme-star-delegation-06

[Russ Housley via Datatracker <[e-mail]>]

Reviewer: Russ Housley
Review result: Not Ready

I reviewed this document as part of the Security Directorate's ongoing
effort to review all IETF documents being processed by the IESG.  These
comments were written primarily for the benefit of the Security Area
Directors.  Document authors, document editors, and WG chairs should
treat these comments just like any other IETF Last Call comments.

Document: draft-ietf-acme-star-delegation-06
Reviewer: Russ Housley
Review Date: 2021-03-14
IETF LC End Date: 2021-03-22
IESG Telechat date: unknown


Summary: Not Ready


Major Concerns:

Abstract: It says: "...  party access to a certificate associated with
said identifier."  This is odd wording, and it is incorrect.  The party
needs access to the private key that corresponds to the public key in
the certificate, and the certificate needs to contain the subject for
"said identifier".  Clearly, all of that should not go in the Abstract,
but what does appear in the Abstract needs to be technically accurate.

Section 1 says: "...   name matches the authority ...".  I find this
description confusing.  I think it would be more clear to say that the
cache server needs to present a certificate whose subject name matches
the domain name of the URL that is requested.  The current wording is
very easy to confuse name of the Certification Authority.

Section 1 says:

   While the primary use case we address is delegation of STAR
   certificates, the mechanism proposed here accommodates any
   certificate managed with the ACME protocol.  See Section 2.4 for
   details.

This is not much of a hint that long-term certificates are supported
in addition to STAR certificates.  Further, a hint about the handling
of revocation is appropriate here.  Support for long-lived certificates
is in conflict with the title of the document.  Please adjust the title
of the document accordingly.

Section 2.3.2 says: "Besides, when delegation is for a STAR certificate,
..." (in four places in this section).  I find this part of the document
structure a bit confusing.  Maybe it is the lask ot adequate warning
about support for long-lived certificates.  Maybe it the the mixing of
STAR certificate and long-lived certificate processing in one section.
I suggest that separate sections be used to present STAR certificate and
long-lived certificate processing

In Section 2.3.4, the text is begging for one more sentence.  Please
say something about the fact that the STAR certificate will expire
shortly after the automatic renewal process is stopped by the IdO.

Section 2.4 is not sufficient to explain the revocation processing.
Only the NDC has the private key needed to make the ACME revocation
request, but this does not get stated in the text.  Also, it is not
clear to me how the NDC knows where to send the revocation request
since the IdO is the ACME account owner.  In addition, the phrase
"would create a self-inflicted DoS" needs more explanation.

Section 5.6 registers a string name for each extendedKeyUsage OID.
There should be a way to provide the OID in dotted decimal format
as well.  New OIDs are being assigned all the time, and some of them
may not be registered with IANA.

Section 5.6 registers a string name for each type of subjectAltName.
This include otherName, which are identified by an OID.  New OIDs are
being assigned all the time.  For example, draft-ietf-anima-autonomic-
control-plane-30 creates a new otherName.  There should be a way to
provide the the otherName OID in dotted decimal format as well.


Minor Concerns:

Abstract: Please spell out ACME, CDN, and STAR.  These are not marked
as "well known" in the RFC Editor abbreviation expansion list.

Section 1 describes [I-D.mglt-lurk-tls13] as an ongoing effort.  This
is not accurate.  The LURK BoF did not lead to a WG or an effort in an
existing WG.  I think the best way forward is to drop this reference.

Section 1.1: Please change CA to "Certification Authority".  See
Section 3 of RFC 5280.  This changes is also needed elsewhere in the
document.

Section 1.1: Please add CDNI, uCDN, dCDN, PASSPorT, CSR and FQDN
to the list of terms.


Nits:

Section 2 says: "... in this draft ...".  Please use a work that will
still be appropriate when this document becomes an RFC.

Section 2.4: s/Sec. 7.6/Section 7.6/  (and many other places)

IDnits reports:


  ** There are 3 instances of too long lines in the document, the
     longest one  being 4 characters in excess of 72.

  == There are 4 instances of lines with non-RFC6890-compliant IPv4
     addresses in the document.  If these are example addresses, they
     should be changed.

[I suspect these are not IPv4 addresses, but OIDs in dotted decimal.]