Re: [Lsr] more feedback on draft-ietf-lsr-isis-srv6-extensions-04

[Peter Psenak <[e-mail]>]

Hi Chris,

On 05/02/2020 00:27, Chris Bowers wrote:
> LSR,
> 
> I have some more feedback on draft-ietf-lsr-isis-srv6-extensions-04 that 
> I am putting in a separate thread so as not to confuse the other thread 
> related to N and A flags.
> 
> =======
> The end of Section 5 points out several issues that can result in 
> forwarding not working correctly.  The reader might think that the next 
> section is going to discuss protocol mechanisms to avoid these issues.  
> Since this is not the case, I think it would be helpful to add some text 
> near the end of Section 5 like:
> 
> "In order to ensure correct forwarding, network operators should take 
> steps to make sure that this requirement is not compromised."

##PP
sure.

> 
> 
> =========
> 
> In section 6, I think it would be useful to explicitly state the 
> following requirement for SRv6 Locator TLVs and their associated SRv6 SIDs:
> 
> 
> "When anycast SRv6 Locator TLVs for the same prefix are advertised by 
> different nodes, the SRv6 Locator TLVs MUST all advertise identical sets 
> of SRv6 SIDs."

##PP
here's the proposed text:

All the nodes advertising the same anycast locator MUST instantiate the 
exact same set of SIDs under such anycast locator.

> 
> 
> Section 3.3 of RFC 8402 has similar text: "Within an anycast group, all 
> routers in an SR domain MUST advertise the same prefix with the same SID 
> value."  That text only refers to a single SID value, so it seems 
> somewhat open to interpretation text in the context of an SRv6 locator 
> that carries multiple SRv6 SIDs. I think it would be better to avoid any 
> potential ambiguity by using the text proposed above in this document.
> 
> =========
> 
> In section 12.1.2. "Revised sub-TLV table" it might avoid an extra 
> interaction with IANA to add a line for the flex-algo prefix metric 
> (currently 6) indicating "n" for TLV#27.

##PP
flex-algo prefix metric is not defined in this draft, so I don't believe 
we can mention it here.

thanks,
Peter

> 
> ==========
> 
> Thanks,
> 
> Chris
> 
> 
>