Re: [Idr] 2 week WG LC for draft-ietf-idr-shutdown-02 (1/17 to 1/31/2017)

[Job Snijders <[e-mail]>]

BTW, regarding last call & IDR implementation requirements, i can
immediately deliver at least two implementations that also accomodate
the 'reset' case so from that perspective there will be no blocker to
proceed.

Kind regards,

Job

On Sat, Jan 28, 2017 at 01:02:17PM +0100, Job Snijders wrote:
> Hi group,
> 
> On Fri, Jan 27, 2017 at 09:02:05PM -0500, Jeffrey Haas wrote:
> > On Fri, Jan 27, 2017 at 03:21:19PM -0800, Matthew Walster wrote:
> > > On 27 January 2017 at 13:38, Jeffrey Haas <[e-mail]> wrote:
> > > 
> > > > I believe the draft should cover both administrative sub-codes.
> > > > [...] Suggestions of other cases to expand the text-form of the
> > > > notification to other sub-codes.
> > 
> > > There is obvious scope-creep at this point, I would like to suggest
> > > either the draft considered as-is or to allow all Cease sub-codes to
> > > have shutdown communication possible.
> > 
> > I certainly hadn't intended to try to push for a more general widening
> > of the scope; just point out something under the existing use case.
> > 
> > While you give three other examples of places that could benefit from
> > additional data regarding why the session has been sent a CEASE, I
> > don't think all of them are equally clear with respect to that data
> > being text.
> > 
> > As an example, the max prefixes might benefit from a simple tuple.
> > E.g. AFI,SAFI,<limit>. Although keeping this within even a
> > light-weight TLV would permit space for the machine-readable portion
> > along with a human readable one.
> > 
> > I would suggest in the absence of a strong case for text strings in
> > the other sub-codes that for now the draft stay restricted to the
> > administrative reset cases. As this draft proves, if there's consenus
> > to add additional DATA to one of the code points, it can move fast.
> 
> I think Jeff's suggestion to add the Administrative Reset (subcode 4)
> case is a good one. There can be value in attachting contact information
> or a case identifier to BGP resets. Matthew mentioned (outside of this
> mailing list) that this information would especially be useful if after
> the reset, the BGP session does not re-establish. Also, conceptually the
> 'reset' and 'shutdown' are close to each other as they are initiated by
> the Administrator.
> 
> However, I consider the other subcodes out of scope for what
> draft-ietf-idr-shutdown tries to accomplish, and as such would belong in
> their own Internet-Draft.
> 
> Kind regards,
> 
> Job