Re: [OSPF] AD review of draft-ietf-ospf-rfc4970bis-02

Acee

Alia having alerted me to the IANA Considerations :-( two further
thoughts.

Jeff Haas has written an excellent I-D recommending that the first and
last points of a numeric registry should be reserved, which I agree
with. In that light, I suggest reserving 8191 in OSPFv3 LSA Function
Codes

Second, 5226bis proposes a cleaner terminology, replacing top-level
registry, registry, sub-registry, sub-sub registry etc with Registry
within Group.  (OSPF has eleven Groups, BGP but one).  I see that OSPFv3
LSA Function Codes is in the Open Shortest Path First v3 (OSPFv3)
Parameters Group.  I think that you are implying that IANA should update
some but not all the references for that Registry.  Thus the reference
for the Registry as a whole should be for this new I-D but the reference
for individual entries for the most part do not change except where they
are to RFC4970(?).

Likewise for OSPF RI TLVs  which is in the Open Shortest Path First v2
(OSPFv2) Parameters Group, some references will need updating, others
will not, and I think that that needs specifying in the I-D.

OSPF Router Informational Capability Bits I think is another Registry
within the same Group while OSPF Router Functional Capability Bits I
think is a new Registry; I don't know which Group it fits best with but
I think that that should be specified.

Tom Petch

----- Original Message -----
From: "Acee Lindem (acee)" <acee@cisco.com>
To: "Alia Atlas" <akatlas@gmail.com>; "OSPF List" <ospf@ietf.org>;
<draft-ietf-ospf-rfc4970bis@ietf.org>
Sent: Wednesday, September 23, 2015 3:37 PM
Subject: Re: [OSPF] AD review of draft-ietf-ospf-rfc4970bis-02

> Hi Alia,
>
> From: Alia Atlas <akatlas@gmail.com<mailto:akatlas@gmail.com>>
> Date: Tuesday, September 22, 2015 at 6:00 PM
> To: OSPF WG List <ospf@ietf.org<mailto:ospf@ietf.org>>,
"draft-ietf-ospf-rfc4970bis@ietf.org<mailto:draft-ietf-ospf-rfc4970bis@i
etf.org>"
<draft-ietf-ospf-rfc4970bis@ietf.org<mailto:draft-ietf-ospf-rfc4970bis@i
etf.org>>
> Subject: AD review of draft-ietf-ospf-rfc4970bis-02
>
> As is customary, I have done my AD review of
draft-ietf-ospf-rfc4970bis-02.  First, let me thank Acee for his work on
this draft.
>
> I have two major concerns before asking for an IETF Last Call.  I
would like to have them
> resolved by this Thursday so that the draft can make the Oct 15 IESG
telechat.
>
> First, from a process concern, I do not see any active discussion on
the OSPF mailing list - even to simply say "yes - go forward".  I don't
see anything about this draft or discussion in minutes for IETF 92 or
IETF 93.   I'd prefer some indication besides silence and lack of
opposition.  I do realize that there are some process or
protocol-tidying drafts where there isn't
> much interest.  However, I am particularly concerned because this is
changing RFC4970 is a way that should be backwards compatible but might
trigger issues.   I would encourage WG participants to PLEASE RESPOND!
>
> Second, I can see the intent that by creating an Opaque ID and
creating a special meaning for 0, the draft is making efforts to
preserve backwards compatibility.  Please add a paragraph or subsection
that articulates how and why backwards compatibility isn't an issue.
>
> I will add more on this.
>
>   For extra credit, what happens if the same TLV information is
advertised in multiple RI LSAs (as part of moving it from one RI LSA to
another)?
>
> This depends on the context of the TLV. In some cases the router
information is additive and in others conflicts need to be resolved.
>
>
>
> Are there any implementations of this draft?  Has backwards
compatibility been verified at all?
>
> Many implementations of RFC 4970. I don’t know of any implementations
that originate multiple RI LSAs. However, with all the proposed OSPF RI
LSA usages, I can see them coming.
>
>
>
> My minor issue is around the IANA considerations; I have detailed
comments below.
>
> Here are additional comments.
>
> 1) In Sec 2: "The first Opaque ID, i.e., 0, should always contain the
Router
>    Informational Capabilities TLV and, if advertised, the Router
>    Functional Capabilities TLV."  and Sec 2.2 "The first instance ID,
i.e., 0,
>    should always contain the Router Informational Capabilities TLV
and,
>    if advertised, the Router Functional Capabilities TLV."
>
>    Since I assume this is important for backwards compatibility,
should those
>    be SHOULD instead of should?
>
> Yes.
>
>
>
>
> 2) In Sec 2.3: "The first defined TLV in the body of an RI LSA is the
Router
>    Informational Capabilities TLV."
>
>    Surely that is only for the first Opaque ID=0?  Does each
subsequent RI LSA
>    also need to contain a Router Informational Capabilities TLV??
>
> No. Will clarify this.
>
>
>
> 3) In Sec 4 IANA Considerations:  This section is defining the
different IANA policies;
> when RFC4970 was written, RFC5226 didn't exist.  But since you're
doing a bis,
> perhaps you can align to the policies in RFC5226 and remove the
unnecessary text??
>
> Yes. I think of already done this.
>
>
> 4) In Sec 4 IANA Considerations:  The registry for OSPFv3 LSA Function
Codes can
> be found at
http://www.iana.org/assignments/ospfv3-parameters/ospfv3-parameters.xhtm
l#ospfv3-parameters-3 and what is in the draft doesn't match up.  I'd
settle for defining the ranges, and value 12 - but it's up to you and
IANA on the preferences.  Would it make sense to change the policy from
Standards Action to IETF Review here also?
>
> Yes - I think this draft should change the policy for OSPFv3 LSA
Functions codes since RFC 4970 initially created the registry. If were
going to change the policy (and I think we want to do this), it makes
sense to do it here from a continuity standpoint.
>
>
>
> Same thing applies to the OSPF RI TLVS
(http://www.iana.org/assignments/ospfv2-parameters/ospfv2-parameters.xht
ml#ospfv2-parameters-9 )   Also here, I think there
> was agreement among the 4 of us who commented on the WG mailing list
to change this
> from Standards Action to IETF Review.
>
> Yes.
>
>
> 5) In Sec 4 IANA Considerations: "All Router Functional Capability TLV
>        additions are to be assigned through standards action."   Given
the discussion
> about IETF Review vs. Standards Action for other registries, are you
sure you want
> Standards Action?
>
> No  - I will change this.
>
>
> I'm sure that we'll get this moving along quickly.
>
> Great - will work on this today in preparation for the tomorrow’s
telechat.
>
> Thanks,
> Acee
>
>
>
> Thanks again!
> Alia
>

------------------------------------------------------------------------
--------

> _______________________________________________
> OSPF mailing list
> OSPF@ietf.org
> https://www.ietf.org/mailman/listinfo/ospf
>