Re: [Isis-wg] WG Last Call for draft-ietf-isis-auto-conf-04

Les 

On 1/22/17, 7:36 PM, "Les Ginsberg (ginsberg)" <ginsberg@cisco.com> wrote:

>Acee -
>
>> -----Original Message-----
>> From: Acee Lindem (acee)
>> Sent: Sunday, January 22, 2017 3:56 PM
>> To: Les Ginsberg (ginsberg); Christian Hopps; isis-wg@ietf.org
>> Cc: draft-ietf-isis-auto-conf@ietf.org; isis-chairs@ietf.org;
>>isis-ads@ietf.org
>> Subject: Re: [Isis-wg] WG Last Call for draft-ietf-isis-auto-conf-04
>> 
>> Hi Les,
>> 
>> On 1/22/17, 12:57 PM, "Les Ginsberg (ginsberg)" <ginsberg@cisco.com>
>> wrote:
>> 
>> >Acee -
>> >
>> >Thanx for reviewing the document.
>> >Responses inline.
>> >
>> >> -----Original Message-----
>> >> From: Isis-wg [mailto:isis-wg-bounces@ietf.org] On Behalf Of Acee
>> >>Lindem
>> >> (acee)
>> >> Sent: Sunday, January 22, 2017 5:39 AM
>> >> To: Christian Hopps; isis-wg@ietf.org
>> >> Cc: draft-ietf-isis-auto-conf@ietf.org; isis-chairs@ietf.org;
>> >>isis-ads@ietf.org
>> >> Subject: Re: [Isis-wg] WG Last Call for draft-ietf-isis-auto-conf-04
>> >>
>> >> Hi IS-IS WG,
>> >>
>> >> I have reviewed the document and support publication. I have the
>> >>following  minor comments:
>> >>
>> >>     1. It should be made clear that the A-Bit indicates that an IS-IS
>> >>router  supports auto-configuration and, is not, necessarily
>> >>auto-configured itself.
>> >> After reading the whole draft, I know that this is the definition of
>> >>the bit but  the initial text says the router is ³operating in
>> >>auto-configuration mode.²
>> >
>> >[Les:] It is clearly stated that the A flag does indeed mean
>> >
>> >" the router is operating in auto-configuration mode."
>> >
>> >I do not see any text which suggests otherwise.
>> 
>> But there is no prior definition of “auto-configuration mode”. I think
>>most
>> readers would believe that this indicates that only routers performing
>>auto-
>> configuration will form adjacencies. Yet the documents
>> states:
>> 
>>    This document also defines mechanisms to prevent the unintentional
>>    interoperation of auto-configured routers with non-autoconfigured
>>    routers.  See Section 3.3.
>> 
>> 
>> Where is the interoperation? This definitely needs to be clarified - I
>>don’t see
>> how the authors can argue on this point!
>[Les:] Section 3.4.2.  Adjacency Formation
>
> "  Routers operating in auto-configuration mode MUST NOT form
>   adjacencies with routers which are NOT operating in auto-
>   configuration mode.  The presence of the Router Fingerprint TLV with
>   the A bit set indicates the router is operating in auto-configuration
>   mode."
>
>I do not see that anything further is needed.
>??

The problem is that there is no prior definition of auto-configuration
mode. Hence, one could interpret the above text as only allowing
adjacencies between auto-configured routers. This directly contradicts the
interoperability with non-autoconfigured routers.

>
>> 
>> >
>> >???
>> >
>> >>     2. In the duplicate detection in section 3.4.3, could you note
>> >>that an IS-IS  router should be able to detect discern the case where
>> >>two interfaces on the  IS-IS router performing auto-configuration are
>> >>connected to the same  network.
>> >>
>> >[Les:] Multiple connections of the same system to the same network can
>> >occur in the absence of auto-configuration and detection of this case
>> >is not altered by auto-configuration. This is detected by receiving a
>> >hello with the same source MAC address as a local interface. There are
>> >then the following cases:
>> >
>> >1)Two interfaces on the local router are connected to the same media.
>> >This is further validated by having the same systemID. The means for
>> >detecting this as well as resolving this are not altered by
>> >auto-configuration.
>> >
>> >2)Two neighbors connected to the same network have the same source
>> MAC
>> >address. This is distinguished by having different system IDs in the
>> >hellos. The means for detecting this as well as resolving this are not
>> >altered by auto-configuration.
>> >
>> >3)Two neighbors connected to the same network have the same source
>> MAC
>> >address and the same systemID. This is distinguished by having
>> >different router fingerprint TLVs in the hellos - something only an
>> >auto-config router could do. But the additional detection capability
>> >does not provide any additional means to correct this issue.
>> >
>> >The authors discussed this point during the writing of the draft and
>> >decided specifically NOT to comment on this issue as it by nature is no
>> >different than what can occur without auto-config and there is no good
>> >way to automatically recover from this case i.e. clearly we cannot
>> >alter the physical connections by programmatic means - nor do we
>> >assume/require a programmatic capability of assigning MAC addresses.
>> >
>> >So, I am not sure what we could say other than to note that this can
>> >occur - but non-auto-config implementations already have to detect this
>> >- so does it make sense to comment on this in the auto-config draft?
>> 
>> Given that consequences of this mis-wiring are more severe when IS-IS
>>auto-
>> configuration is being used, I think this deserves at least the
>>discussion above
>> included in the draft.
>> 
>[Les:] I do not see that this issue is any more/less severe when
>operating in autoconfig mode.
>Manual intervention is required to resolve the issue regardless of mode -
>the protocol cannot heal itself in this case. All we can do is send out a
>notification and be smart in the implementation so as to avoid constant
>adjacency churn. This behavior is required/recommended regardless of
>autoconfig mode. In fact, it could be argued the problem is more severe
>for larger networks as the side effects of churn associated with
>sub-optimal handling of this problem will be far worse in a large network.
>
>Interestingly, I do not even see a notification defined for this
>condition in the MIB (RFC 4444) - perhaps we will do better when defining
>the YANG data model. :-)
>
>This is perhaps a problem worth discussing - but I don’t see that it is
>in any way unique to or related to autoconfig - so adding it to this
>specification doesn’t seem appropriate.

With the respect to this being a problem in any mode, I agree. One
motivation for at least having a statement in this draft is that it
specifically addresses duplicate System-ID handling and I’m not sure
whether that is addressed in any other IS-IS specifications. However, I
don’t feel that strongly. Here is what we had in RFC 7503:

 An OSPFv3 router implementing this specification should ensure that
 the inadvertent connection of multiple router interfaces to the same
 physical link is not misconstrued as detection of an OSPFv3 neighbor
 with a duplicate Router ID.

Thanks,
Acee 

>
>   Les
>
>
>> Thanks,
>> Acee
>> 
>> 
>> 
>> >
>> >   Les
>> >
>> >
>> >> Thanks,
>> >> Acee
>> >>
>> >> On 1/17/17, 7:00 AM, "Isis-wg on behalf of Christian Hopps"
>> >> <isis-wg-bounces@ietf.org on behalf of chopps@chopps.org> wrote:
>> >>
>> >> >
>> >> >Hi Folks,
>> >> >
>> >> >We are starting a WG Last Call for
>> >> >
>> >> >  "ISIS Auto-Configuration"
>> >> >  - https://datatracker.ietf.org/doc/draft-ietf-isis-auto-conf/
>> >> >
>> >> >The WGLC will expire in 2 weeks on Jan 31, 2017.
>> >> >
>> >> >Thanks,
>> >> >Chris & Hannes.
>> >>
>> >> _______________________________________________
>> >> Isis-wg mailing list
>> >> Isis-wg@ietf.org
>> >> https://www.ietf.org/mailman/listinfo/isis-wg
>