Re: [Lsr] 答复: Working Group Adoption of "IGP Unreachable Prefix Announcement" - draft-ppsenak-lsr-igp-ureach-prefix-announce-04

[Aijun Wang <[e-mail]>]

Hi, Peter:

Aijun Wang

China Telecom

On Sep 6, 2023, at 15:57, Peter Psenak <ppsenak=[e-mail]> wrote:

Aijun,

WG adoption should be done based on the draft content, the quality of the solution it describes and not based on the draft age or order.

[WAJ] Not exactly. We should also respect the original contributions.

Multiple people have pointed out over the years that the solution that you propose in your draft - e.g. using router-id of 0 to indicate the unreachability is broken and lacks the backward compatibility aspect. Your original draft did NOT include the unreachable metric and even though you added it later (way after it was proposed in the other draft), your draft still uses the original router-id 0 idea. The fact that you need the PUAM Capabilities Announcement in your draft speaks for itself.

[WAJ] The reason that we keep the “PUAM Capabilities Announcement” is that we want to fade-out the usage of “LSInfinity” Metric within the network,  as described in the draft:

If not all of nodes within one area support the PUAM capabilities, the PUAM message should be advertised with the associated prefix cost set to LSInfinity. According to the description in [https://www.rfc-editor.org/info/rfc2328" class="cite xref" rel="nofollow">RFC2328], [https://www.rfc-editor.org/info/rfc5305" class="cite xref" rel="nofollow">RFC5305] and [https://www.rfc-editor.org/info/rfc5308" class="cite xref" rel="nofollow">RFC5308], the prefix advertised with such metric value will not be considered during the normal SPF computation, then such additional information will avoid the misbehavior of the nodes when they don't recognize the PUAM message.

If all of nodes within one area support the PUAM capabilities, the PUAM message can be safely advertised without the additional LSInfinity metric information.

In the latest version of your draft, you have stated also that there maybe other cases to use such metrics:

Even though in all cases the treatment of such metric, or NU-bit, is specified for IS-IS, OSPF and OSPFv3, having an explicit way to signal that the prefix was advertised in order to signal unreachability is required to distinguish it from other cases where the prefix with such metric is advertised.

In your proposal, the LSInfinity metric must be set even all the nodes within the area have the capabilities to understand the explicit signaling mechanism. There will be more confusions when other cases of LSInfinity usage proposed.

As explained previously, set the existing original router id to NULL is one simple, uniform way to express unreachable information for OSPFv2/v3 and IS-IS. We needn’t find the different places to define and flag it for different protocol. 

Isn’t it more easier for the implementation?

It’s also straightforward to understand.

Though you may have been the first one to try to solve the problem in the IETF draft, your solution is still incorrect. As a result of your inability to listen to the comments an alternative draft was written that is technically superior, backward compatible, has wide support from vendors as well as operators, including ones that were originally co-authors on your draft and decided to join the alternate draft based on its technical advantages, has been implemented and deployed in the field.

[WAJ]There are other reasons for the switch over of co-author, I think we needn’t expand or refer to it. The standard is not the technical implementation of one company. The standard should look forward further the scenarios, deployment considerations and solutions coverages.

As a matter of fact, I have invited you to join our draft several times, but you refused. You insisted on pushing your draft.

[WAJ] I have proposed several times the merger of two drafts, we can discuss the contents and structure of the converged document. It’s pity that we missed the IETF117 on-site meeting. 

But it’s also hurry to put forward only your draft when we consider the history, contributions and similarities of two proposals.

My proposal is that we postpone the adoption of this draft, and discuss offline the merger document on the coming IETF118 meeting. 

Then, we can submit the merged document to the LSR WG for adoption call.

my 2c,
Peter





On 06/09/2023 07:56, Aijun Wang wrote:

Hi, Acee:

AGAIN, before making some assertions, please check the following fact:

Have you noticed the 00 version of https://datatracker.ietf.org/doc/draft-ppsenak-lsr-igp-event-notification/ <https://datatracker.ietf.org/doc/draft-ppsenak-lsr-igp-event-notification/> was submitted on July 5, 2021?

But the description about the short lived notification in https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-06#section-7 <https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-06#section-7> was on March 26, 2021.

Then, which draft was the first?

For the adoption call or merge efforts, I think the WG should consider the following facts:

1)Which draft is the first to provide the use cases?

2)Which draft is the first to propose explicit signaling for unreachable information?

3)Which draft is the first to propose short lived notification?

4)Which explicit signaling mechanism is simpler?

5)Which draft provides more mechanisms to cover more scenarios?

The base document should be selected based on the answers of the above questions.

Select the base document doesn’t mean that it can’t be changed before the adoption(I haven’t said “Without Change” is the merge condition).

Actually, we welcome more authors to join us to finalize the document and solution.

As one of the most important WG within IETF, I think LSR WG should respect the original contributions to the WG.

It is too hurry to consider or adopt only the draft that you prefer, especially the follower draft.

Best Regards

Aijun Wang

China Telecom

*发件人:*[e-mail] [mailto:[e-mail]] *代表 *Acee Lindem

*发送时间:*2023年9月6日0:56

*收件人:*Aijun Wang <[e-mail]>

*抄送:*Robert Raszuk <[e-mail]>; Les Ginsberg (ginsberg) <ginsberg=[e-mail]>; Huzhibo <huzhibo=[e-mail]>; Peter Psenak <[e-mail]>; linchangwang <[e-mail]>; lsr <[e-mail]>

*主题:*Re: [Lsr] Working Group Adoption of "IGP Unreachable Prefix Announcement" - draft-ppsenak-lsr-igp-ureach-prefix-announce-04

Hi Aijun,

When the WG discussion first indicated that this was a use case that needed to be addressed, I don’t dispute that you immediately added it to your draft.

I have no doubt you would have purported support of any use case under discussion.

However, the first draft to address this use case with a short-lived notification was https://datatracker.ietf.org/doc/draft-ppsenak-lsr-igp-event-notification/ <https://datatracker.ietf.org/doc/draft-ppsenak-lsr-igp-event-notification/>

Based on WG feedback and collaboration of multiple vendors, this draft evolved to draft-ppsenak-lsr-igp-ureach-prefix-announce.

While you’ve incorporated elements of the draft under discussion, your draft still includes pieces (sometimes conflicting) from previous use cases.

There was an effort to merge the drafts but you declined unless your draft was used (without change) as the base. I’m not sure your motivation.

Thanks,

Acee

   On Sep 1, 2023, at 20:25, Aijun Wang <[e-mail]

   <mailto:[e-mail]>> wrote:

   Hi, Acee:

   Act as LSR chair, I think you should be more responsible to make any

   unfounded assertions:

   We have described the previous statements in

   https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-06#section-7 <https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-06#section-7>, March 26, 2021, one year before the 00 version of draft-ppsenak(March 25,2022)

   Then, which draft copy or incorporate which draft?

   Aijun Wang

   China Telecom

       On Sep 1, 2023, at 20:05, Acee Lindem <[e-mail]

       <mailto:[e-mail]>> wrote:

       Hi Aijun,

           On Aug 31, 2023, at 23:36, Aijun Wang

           <[e-mail]

           <mailto:[e-mail]>> wrote:

           Hi，Acee:

           Please

           readhttps://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-12#section-7 <https://datatracker.ietf.org/doc/html/draft-wang-lsr-prefix-unreachable-annoucement-12#section-7>before making misguide assertions:

           “The advertisement of PUAM message should only last one

           configurable period to allow the services that run on the

           failure prefixes are switchovered.”

       I guess I haven’t kept up with all the elements of the draft

       under adoption that you continue to incorporate into your draft.

       This has been a continuing theme since initial discussed of the

       application signaling use case. While I have no interest in

       improving your draft, making the LSP/LSA short-lived conflicts

       with the other scenarios your draft purports to address.

       Acee

           Best Regards

           Aijun Wang

           China Telecom

           *发件人:*[e-mail]

           <mailto:[e-mail]>[mailto:[e-mail]

           <mailto:[e-mail]>]*代表*Acee Lindem

           *发送时间:*2023年9月1日0:50

           *收件人:*Robert Raszuk <[e-mail]

           <mailto:[e-mail]>>

           *抄送:*Les Ginsberg (ginsberg)

           <ginsberg=[e-mail]

           <mailto:ginsberg=[e-mail]>>; Huzhibo

           <huzhibo=[e-mail]

           <mailto:huzhibo=[e-mail]>>; Peter Psenak

           <[e-mail] <mailto:[e-mail]>>; linchangwang

           <[e-mail]

           <mailto:[e-mail]>>; lsr <[e-mail]

           <mailto:[e-mail]>>

           *主题:*Re: [Lsr] Working Group Adoption of "IGP Unreachable

           Prefix Announcement" -

           draft-ppsenak-lsr-igp-ureach-prefix-announce-04

               On Aug 31, 2023, at 12:32, Robert Raszuk

               <[e-mail] <mailto:[e-mail]>> wrote:

               Hi Acee,

                   In any case, one will need to update the signaling

                   routers and the routers acting on the signal.

               I guess this is clear to all.

                   Additionally, your request for the adoption was that

                   the draft have a stronger statement about the

                   mechanism being used for solely for signaling for

                   applications (e.g., BGP PIC).

               As to the applicability my comment was that either draft

               should state in strong normative language that this is

               applicable only to applications which data plane uses

               encapsulation to the next hop.

               Said this draft-wang introduces the additional

               signalling, sort of trying to assure that all nodes in

               an area understand the new messages - but I am not sure

               if even advertising PUAM capability means that it will

               be actually used for all destinations ?

           No - but while the draft under adoption (ppsenak-lsr…) is

           for an ephemeral signal which the WG agreed was a valid use

           case, in the other draft, the LSAs are long-lived and are

           also may be used for other purposed than signaling (e.g.,

           reread both sections 4 and 6 of draft-wang-lsr…). This draft

           starting with a whole different use case but selectively

           added mechanisms from ppsenak-lsr…

           I seem to recall you were a strong proponent of limiting the

           scope.

                   By responding to this Email inline, some may believe

                   you support the assertion that we should start the

                   adoption of both drafts. Please be clarify this.

               Well the way I see this is that adoption call is a bit

               more formal opportunity for WG members to express their

               opinion on any document. But maybe LSR (for good

               reasons) have different internal rules to decide which

               document should be subject to WG adoption and does sort

               of pre-filtering.

               If adoption call proves document has negative comments

               or lacks cross vendor support it simply does not get

               adopted.

               Maybe I am just spoiled looking at how IDR WG

               process works :-)

           You replied to an Email inline suggesting adoption of both

           drafts. That is what I think could have been misconstrued -

           especially by those who didn’t follow the discussion until

           now who think you are agreeing with this recommendation.

                   As for your other comment that this could be

                   accomplished with BGP or an out-of-bound mechanism,

                   that is true but that could be true of many problem.

                   However, the solution under adoption has running

                   code and wide vendor support.

                 Right ... As I wrote to Peter - perhaps this is just a

               pragmatic approach and flooding is what link state uses

               so be it.

               As you know I did try in the past to propose BGP

               Aggregate withdraw but then feedback of the community

               was that PEs do not go down that often to justify the

               extension.

           Hmm…We seem to have broad support for the LSR application

           signaling use case.

           Thanks,

           Acee

               Best,

               Robert

           _______________________________________________

           Lsr mailing list

           [e-mail] <mailto:[e-mail]>

           https://www.ietf.org/mailman/listinfo/lsr

           <https://www.ietf.org/mailman/listinfo/lsr>

       _______________________________________________

       Lsr mailing list

       [e-mail] <mailto:[e-mail]>

       https://www.ietf.org/mailman/listinfo/lsr

       <https://www.ietf.org/mailman/listinfo/lsr>

   _______________________________________________

   Lsr mailing list

   [e-mail] <mailto:[e-mail]>

   https://www.ietf.org/mailman/listinfo/lsr

   <https://www.ietf.org/mailman/listinfo/lsr>

_______________________________________________
Lsr mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/lsr