Hi Peter, Joe, et al,
On 10/30/18, 8:05 AM, "Peter Psenak (ppsenak)" <ppsenak@xxxxxxxxx> wrote:
Hi Joe,
thanks for your review, please see inline (##PP):
On 26/10/18 21:42 , Joe Clarke wrote:
> Reviewer: Joe Clarke
> Review result: Has Nits
>
> I have been assigned to review
> draft-ietf-ospf-ospfv3-segment-routing-extensions on behalf of the ops
> directorate. This document defines OSPFv3 extensions needed for segment
> routing (SR). And therein lies my first nit. While the document begins to set
> forth this overarching scope, a small paragraph in section 1 further limits it
> to MPLS dataplanes only. I think perhaps the abstract should be updated to
> clarify that.
##PP
Done
> Other items I found are listed below.
>
> Overall, there are a lot of terminology used like RSVP, LDP, LSP, SID, etc. I
> think this document would benefit from a terminology section.
##PP
added
>
> With respect to TLV types 8, 9, 14, and 15, they are defined in
> draft-ietf-ospf-segment-routing-extensions, and it took me a while to figure
> out where you were getting those values and why they weren't spelled out in the
> IANA considerations. You have a normative reference to this, which is good,
> but you only mention it with respect to the algorithm parameters. I think
> another mention is required.
>
> I'm going to be pedantic here. According to RFC7770, when a new OSPF Router
> Information LSA TLV is defined, the spec needs to explicitly state if it's
> applicable to OSPFv2, v3, or both. While you reference the TLVs from
> draft-ietf-ospf-segment-routing-extensions, I didn't see that either document
> _explicitly_ states that they are applicable to both.
##PP
added the following to each of the values:
Type: X as defined in [I-D.ietf-ospf-segment-routing-extensions] and
aplicable to OSPFv3.
>
> ===
>
> Section 2.1
>
> s/length is other then 3 or 4/length is other than 3 or 4/
##PP
fixed
>
> ===
>
> Section 3.2
>
> s/If more then one SID/Label/If more than one SID/label/
##PP
fixed
>
> ===
>
> Section 3.2
>
> "When a router receives multiple overlapping ranges, it MUST
> conform to the procedures defined in
> [I-D.ietf-spring-segment-routing-mpls]."
>
> It would be useful to include a section pointer here. I think your referring
> to Section 2.3 where the router ignores the range? Is it likely that will
> change to something other than "ignore?" If not, maybe it's just worth
> mentioning that here.
##PP
I don't think it is good to specify the behavior which is described
somewhere else. Regarding the section, the
ietf-spring-segment-routing-mpls is still being worked on and the
section may changes. We used the same text in OSPFv2 and ISIS SR drafts.
I would like to be consistent here.
Given that this is a normative reference, I don't think it would create too much of a dependency to include the section in the reference. We've had a protracted discussion (1-2 years) on the whole SID overlap topic in SPRING and I believe we've finally come up with behavior and the specification of such behavior with which everyone agree (or at least doesn't strongly disagree).
>
> ===
>
> Section 3.3
>
> s/If more then one SID/Label/If more than one SID/Label/
##PP
fixed.
>
> ===
>
> Section 3.3
>
> "The originating router MUST NOT advertise overlapping ranges."
>
> You specify what a router should do if it receives overlapping ranges above. I
> think the same text should be used here, too.
##PP
Here we say that the originating router MUST NOT advertise overlapping
ranges. We can not specify what it should do when it breaks the MUST.
We specify what other routers should do when they receive overlapping
ranges and we refer it to spring-segment-routing-mpls draft. Again this
is the same as we used in OSPFv3 and ISIS SR extensions. I would like to
keep the consistency here.
>
> ===
>
> Section 5
>
> "Other bits: Reserved. These MUST be zero when sent and are
> ignored when received."
>
> The normative language changes. In other places you say the bits SHOULD be 0.
> I suggest:
##PP
Whenever we refer to "other bits" in the flag fields we use the same
language.
>
> Other bits: Reserved. These SHOULD be set to 0 when sent and MUST be ignored
> when received.
##PP
this refers to Reserved fields in the TLV (not the bits in a flag field)
and again is used consistently across document.
I agree. Use "These SHOULD be set to 0 when sent and MUST be ignored when received." For all reserved bits.
Thanks,
Acee
>
> ===
>
> Section 7.4.1
>
> s/state lower then 2-Way/state lower than 2-Way/
##PP
fixed.
thanks,
Peter
>
> ===
>
>
> .
>
