Re: [Isis-wg] Alexey Melnikov's Discuss on draft-ietf-isis-rfc4971bis-03: (with DISCUSS and COMMENT)

[Alexey Melnikov <[e-mail]>]

Hi Les,

On Thu, Aug 18, 2016, at 02:18 PM, Les Ginsberg (ginsberg) wrote:
> Alexey -
> 
> The document states in Section 2:
> 
> " The Router CAPABILITY TLV is OPTIONAL.  As specified in Section 3,
>    more than one Router CAPABILITY TLV from the same source MAY be
>    present."
> 
> This clearly allows multiple CAPABILITY TLVs from the same source to be
> present. Why? Because the basic unit of advertisement in IS-IS is a TLV -
> in this case one which is intended to include multiple sub-TLVs. As the
> value field of a  TLV is limited in size to 255 bytes if a router has
> more than 255 bytes of CAPABILITY information to advertise then multiple
> TLVs will be required. This isn't a "trick" - it is fundamental to the
> operation of IS-IS. 

I was actually looking for this text in the document and didn't find it.
Can you just say this in the document?

> It is unclear to me how you could interpret the above text as meaning
> that multiple CAPABILITY TLVs from the same source are either disallowed
> or unexpected.

The text in Section 3:

   Where a receiving system has two copies of a CAPABILITY TLV from the
   same system that have different settings for a given attribute, the
   procedure used to choose which copy shall be used is undefined.

is telling me that in some cases behaviour with multiple CAPABILITY TLVs
is undefined. I was trying to figure out what are the conditions, thus
all my questions in this thread.

So obviously I am missing something. Can you please help me understand
how the two statements are not in conflict?

> I appreciate that you may not be an expert in IS-IS and some fundamentals
> may be unfamiliar to you. That is why the document provides references to
> the base specifications which provide these definitions.  A new
> implementer of the protocol could not write an implementation solely
> based on this document - and it is not reasonable to request that this
> document be written without dependencies on the base specifications.

Of course not. I was trying to get a copy of the base spec, but it looks
like I need to pay for it.

>    Les
> 
> > -----Original Message-----
> > From: Alexey Melnikov [mailto:[e-mail]]
> > Sent: Thursday, August 18, 2016 2:59 AM
> > To: Les Ginsberg (ginsberg); The IESG
> > Cc: [e-mail]; Christian Hopps; [e-mail];
> > [e-mail]
> > Subject: Re: Alexey Melnikov's Discuss on draft-ietf-isis-rfc4971bis-03: (with
> > DISCUSS and COMMENT)
> > 
> > Hi Les,
> > 
> > On Tue, Aug 16, 2016, at 05:56 PM, Les Ginsberg (ginsberg) wrote:
> > > Alexey -
> > 
> >  [snip]
> > > > --------------------------------------------------------------------
> > > > --
> > > > DISCUSS:
> > > > --------------------------------------------------------------------
> > > > --
> > > >
> > > > I would like to get clarification on the following points before
> > > > recommending approval of this document:
> > > >
> > > > 1) How do multiple CAPABILITY TLVs from the same source treated, if
> > > > they have the same S and D flags, but different subTLV? Are the
> > > > cumulative? Or this is not allowed?
> > > > I am sorry if I missed where this was described, let me know if I did.
> > >
> > > [Les:] Section 2 states:
> > >
> > > " The Router CAPABILITY TLV is OPTIONAL.  As specified in Section 3,
> > >    more than one Router CAPABILITY TLV from the same source MAY be
> > >    present."
> > >
> > > The only problematic case is when you have two TLVs from the same
> > > source which have different settings for the same attribute i.e. same
> > > sub-TLV but different values. This case is discussed in detail in Section 3.
> > > Different sub-TLVs presents no issue of course.
> > > Note this text is unchanged from the existing RFC.
> > 
> > What happens in real world if there are two CAPABILITY TLVs which contain
> > different attributes? Are they treated as cumulative (i.e. this is a nice trick to
> > overcome CAPABILITY TLV length limit), does the second CAPABILITY  TLV
> > value always overrides earlier instances of the CAPABILITY TLV?
> > 
> > I think the document should state what happens. If there is no consistent
> > behaviour in this area, the document should says so as well.