Aijun, Speaking as WG member:
At least for OSPF, passive interfaces are not standardized in RFC 2328 or RFC 5340. Hence, this purely a vendor concept. Additionally, it is a property, albeit a vendor property, of a link and not a prefix. It would be both inappropriate and profligate (considering the scarcity) to allocate a prefix option for the purpose of identifying a passive link associated with the prefix. Given your narrow use case of identifying the edge of an IGP domain, it would certainly be better to allocate a new TLV specifically for purpose and perhaps this doesn't belong in the IGPs at all and should be something you propose solely for BGP-LS consumption. Speaking as WG Co-chair: Given strong objections to this draft in its current form, I don't really see a good reason for present it at IETF 109. I believe it would just be a rehash of the discuss that has already taken place. Thanks, Acee On 11/9/20, 4:44 AM, "Peter Psenak" <[email protected]> wrote: Hi Aijun, On 09/11/2020 07:35, Aijun Wang wrote: > Hi, Peter: > > Currently, the inter-AS link TLV is advertised within the Inter-AS-TE-LSA for OSPF and Inter-AS Reachability TLV for ISIS. > But I think these two places are not suitable for the stub-link information. > > It seems that separating the stub-link information from the inter-as link information is better, because not all of the stub-links are inter-as link. > If so, can we put the newly defined Stub-Link TLV within the Router LSA for OSPF and make it one new top TLV for ISIS? Router LSA does not have TLVs, you would have to add the data to Extended Prefix Link TLV (RFC7684), or define a net top-level TLV under the OSPFv2 Extended Link Opaque LSA. For ISIS you don't have a choice really, you need to define a new top-level TLV. thanks, Peter > > > Best Regards > > Aijun Wang > China Telecom > > -----Original Message----- > From: Peter Psenak [mailto:[email protected]] > Sent: Saturday, November 7, 2020 1:56 AM > To: Aijun Wang <[email protected]> > Cc: Aijun Wang <[email protected]>; Acee Lindem (acee) <[email protected]>; [email protected] > Subject: Re: [Lsr] FW: New Version Notification for draft-wang-lsr-passive-interface-attribute-04.txt > > Aijun, > > On 05/11/2020 12:04, Aijun Wang wrote: >> Hi, Peter: >> Then how about defines one new top TLV to flood such information within the IGP? Fox example, Stub-Link TLV? If so, other characteristics associated with the Link can also be advertised accordingly. > > yes, unless you can use or extend the existing inter-AS link advertisement. > > thanks, > Peter > >> >> If acceptable, we can forward this draft along this direction. >> >> >> Aijun Wang >> China Telecom >> >>> On Nov 5, 2020, at 17:15, Peter Psenak <[email protected]> wrote: >>> >>> Aijun, >>> >>> the point I was trying to make was that you should think of a similar mechanism for your use cases - e.g. define something that advertises the link without advertising the IS adjacency and not mess up with the prefix advertisement. >>> >>> thanks, >>> Peter >>> >>>> On 05/11/2020 10:09, Aijun Wang wrote: >>>> Hi, Peter: >>>> Yes, RFC 5392 is the OSPF corresponding part for the inter-AS TE solution. But using these existing solutions has some limitation in deployment, as I explained in https://mailarchive.ietf.org/arch/msg/lsr/VLufuaGDiRgaflcu58FY_SHnJ7A/. >>>> And, in some situations, not all of the passive interfaces are connected with another AS, then flag these interfaces using RFC 5316 or RFC 5392 is not appropriate. >>>> Do you agree? >>>> Best Regards >>>> Aijun Wang >>>> China Telecom >>>> -----Original Message----- >>>> From: [email protected] [mailto:[email protected]] >>>> Sent: Thursday, November 5, 2020 4:26 PM >>>> To: Aijun Wang <[email protected]>; 'Acee Lindem (acee)' >>>> <[email protected]>; 'Aijun Wang' <[email protected]> >>>> Cc: [email protected] >>>> Subject: Re: [Lsr] FW: New Version Notification for >>>> draft-wang-lsr-passive-interface-attribute-04.txt >>>> Hi Aijun, >>>> please look at rfc5316, ISIS already have a way to advertise inter-AS link without forming an adjacency. >>>> thanks, >>>> Peter >>>>> On 05/11/2020 02:15, Aijun Wang wrote: >>>>> Hi, Acee: >>>>> >>>>> Thanks for this comments. >>>>> The consideration for the position of flagging the passive interface have been stated in the updated 05 version https://tools.ietf.org/html/draft-wang-lsr-passive-interface-attribute-05#section-3, as the followings: >>>>> >>>>> ISIS [RFC5029] defines the Link-Attributes Sub-TLV to carry the link >>>>> attribute information, but this Sub-TLV can only be carried within >>>>> the TLV 22, which is used to described the attached neighbor. For >>>>> passive interface, there is no ISIS neighbor, then it is not >>>>> appropriate to use this Sub-TLV to indicate the passive attribute of >>>>> the interface. >>>>> >>>>> OSPFv2[RFC2328] defines link type field within Router LSA, the type 3 >>>>> for connections to a stub network can be used to identified the >>>>> passive interface. But in OSPFv3 [RFC5340], type 3 within the >>>>> Router-LSA has been reserved. The information that associated with >>>>> stub network has been put in the Intra-Area-Prefix-LSAs. >>>>> >>>>> What about your opinions regarding to the above statements? Currently, we think putting the flag within the prefix attribute that associated the passive interface is appropriate. >>>>> If we can find other appropriate/acceptable place to hold this information, we can also update the draft later accordingly. >>>>> >>>>> >>>>> Best Regards >>>>> >>>>> Aijun Wang >>>>> China Telecom >>>>> >>>>> >>>>> -----Original Message----- >>>>> From: Acee Lindem (acee) [mailto:[email protected]] >>>>> Sent: Thursday, November 5, 2020 4:11 AM >>>>> To: Aijun Wang <[email protected]> >>>>> Cc: Aijun Wang <[email protected]>; Peter Psenak (ppsenak) >>>>> <[email protected]>; [email protected] >>>>> Subject: Re: [Lsr] FW: New Version Notification for >>>>> draft-wang-lsr-passive-interface-attribute-04.txt >>>>> >>>>> Hi Aijun, >>>>> You still didn't answer the question as to why you didn't rework this draft for passive interface to be an interface attribute rather than a prefix attribute? >>>>> Thanks, >>>>> Acee >>>>> >>>>>> On 10/1/20, 6:13 AM, "Acee Lindem (acee)" <[email protected]> wrote: >>>>> >>>>> Hi Aijun, >>>>> You didn't answer my question and pruned my message. Other than your attempt to expose the topology of areas outside the area, there is no other reason to associate the passive interface attribute with a prefix. We seem to be in a circular discussion.... >>>>> Acee >>>>> >>>>>> On 9/30/20, 10:43 PM, "[email protected] on behalf of Aijun Wang" <[email protected]> wrote: >>>>> >>>>> Hi, Acee: >>>>> Except the corner cases of unnumbered interface, would you like to illustrate other scenarios that the process does not apply? >>>>> As mentioned in last mail, knowing the passive interfaces can assist the nodes or controller know the boundaries of the network. >>>>> >>>>> Aijun Wang >>>>> China Telecom >>>>> >>>>> > On Sep 30, 2020, at 19:47, Acee Lindem (acee) <[email protected]> wrote: >>>>> > >>>>> >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Lsr mailing list >>>>> [email protected] >>>>> https://www.ietf.org/mailman/listinfo/lsr >>>>> >>>>> >>>> _______________________________________________ >>>> Lsr mailing list >>>> [email protected] >>>> https://www.ietf.org/mailman/listinfo/lsr >>> >> >> >> > > > > _______________________________________________ Lsr mailing list [email protected] https://www.ietf.org/mailman/listinfo/lsr
