Re: [Idr] I-D Action: draft-ietf-idr-error-handling-14.txt

["John G. Scudder" <[e-mail]>]

Rob,

Thanks for your comments. With regard to the first, how do you (and the WG) feel about this paragraph for the end of the Operations section?

   Section 8 mentions that attribute discard should not be used in cases
   where "the attribute in question has or may have an effect on route
   selection."  Although all cases that specify attribute discard in
   this document do not affect route selection by default, in principle
   routing policies could be written which affect selection based on
   such an attribute.  Operators should take care when writing such
   policies to consider the possible consequences of an attribute
   discard.  (In general, as long as such policies are only used on
   external BGP sessions, correctness issues are not expected to arise.)

Regarding ELC, thanks for catching that! Considering that we're excluding deprecated attributes from consideration in this document, and considering that draft-ietf-mpls-deprecate-bgp-entropy-label has been approved for publication as an RFC, I think it's safe for me to remove all references to RFC 6790. 

Regards,

--John

On Oct 16, 2014, at 9:25 AM, Rob Shakir <[e-mail]> wrote:

> Hi John, IDR,
> 
> On 3 Sep 2014, at 20:39, John G. Scudder <[e-mail]> wrote:
> 
>> 
>> In short, it removes the discussion of what it means for an NLRI to be incorrect.
>> Reviewers may want to pay particular attention to S. 7.11.
> 
> Sorry for the delay in sending over some comments.
> 
> First off, thanks to the authors for the work on this doc — it’s taken something that 
> we’ve debated in length on the IDR list and resulted in (IMHO) a very clean document 
> that does a good job of explaining the motivation and revisions to the error handling 
> behaviour.
> 
> I had some very minor comments:
> 
> 	- Currently, when the text discusses attribute discard, it links this to 
> 	  something that cannot affect the route selection. Since in most routing 
> 	  policy languages, one could potentially match on /any/ attribute if one
> 	  so wanted, then I was wondering whether it was worth adding something into
> 	  the operational considerations section that briefly noted that operators
> 	  SHOULD NOT implement policies which influence route selection based on an
> 	  attribute that might be discarded.
> 
> 	- In 7.16, we re-define the error handling for BGP ELC, but I noted the
> 	  existence of draft-ietf-idr-mpls-deprecate-bgp-entropy-label. I’m not sure
> 	  whether it is still worth including this analysis, or whether it is worth
> 	  noting that this attribute is (likely to be) deprecated.
> 
> Other than this, thanks again for the work on this document.
> 
> Kind regards,
> r.