Re: [Lsr] Last Call: <draft-ietf-lsr-isis-srv6-extensions-14.txt> (IS-IS Extension to Support Segment Routing over IPv6 Dataplane) to Proposed Standard

["Acee Lindem (acee)" <[e-mail]>]

Speaking as WG contributor:

From: Lsr <[e-mail]> on behalf of "Les Ginsberg (ginsberg)" <ginsberg=[e-mail]>
Date: Friday, May 7, 2021 at 12:53 PM
To: "Ketan Talaulikar (ketant)" <ketant=[e-mail]>, Alvaro Retana <[e-mail]>, "Peter Psenak (ppsenak)" <[e-mail]>, "[e-mail]" <[e-mail]>
Cc: Christian Hopps <[e-mail]>, "[e-mail]" <[e-mail]>, Gunter Van de Velde <[e-mail]>
Subject: Re: [Lsr] Last Call: <draft-ietf-lsr-isis-srv6-extensions-14.txt> (IS-IS Extension to Support Segment Routing over IPv6 Dataplane) to Proposed Standard

As has been mentioned in this thread, the need for the prefix-attributes sub-TLV to correctly process leaked advertisements is not unique to the Locator TLV. The reason prefix-attributes TLV was created was to address the same gap with IP/IPv6 reachability advertisements.
And I think by now implementations (certainly ones that support newer functionality like SRv6) should have added support for prefix-attributes sub-TLV .

In the case of the Locator TLV  – since this is new functionality – we have the option of mandating prefix-attributes sub-TLV – something we could not do with IP/IPv6 Reachability since that has been deployed for many years.

But,  please recognize two consequences of the MUST option:

1)Implementations may have to deal w  backwards compatibility w early deployments of SRv6. This would only be an issue if there are implementations that currently do NOT send prefix-attributes sub-TLV w Locator TLV.
Are there any such implementations??

2)In the case where the deployment is a single level, it could be argued that prefix-attributes sub-TLV isn’t needed.
I personally would NOT make such an argument, but we should understand that MUST applies to this case as well.

If everyone is OK with these consequences (personally I am OK) then I think it is fine to go with MUST.

I’m fine with these consequences. Better to fix this now and not use the SRv6 Prefix non-deterministically.

Thanks,
Acee

   Les


From: Lsr <[e-mail]> On Behalf Of Ketan Talaulikar (ketant)
Sent: Friday, May 07, 2021 7:00 AM
To: Alvaro Retana <[e-mail]>; Peter Psenak (ppsenak) <[e-mail]>; [e-mail]
Cc: [e-mail]; [e-mail]; Van De Velde, Gunter (Nokia - BE/Antwerp) <[e-mail]>
Subject: Re: [Lsr] Last Call: <draft-ietf-lsr-isis-srv6-extensions-14.txt> (IS-IS Extension to Support Segment Routing over IPv6 Dataplane) to Proposed Standard

Hi Peter,

I agree that the support for the Prefix Attribute Flags TLV is required in the Locator TLV.

Thanks,
Ketan

From: Lsr <[e-mail]<mailto:[e-mail]>> On Behalf Of Alvaro Retana
Sent: 07 May 2021 19:23
To: Peter Psenak (ppsenak) <[e-mail]<mailto:[e-mail]>>; [e-mail]<mailto:[e-mail]>
Cc: [e-mail]<mailto:[e-mail]>; [e-mail]<mailto:[e-mail]>; Van De Velde, Gunter (Nokia - BE/Antwerp) <[e-mail]<mailto:[e-mail]>>
Subject: Re: [Lsr] Last Call: <draft-ietf-lsr-isis-srv6-extensions-14.txt> (IS-IS Extension to Support Segment Routing over IPv6 Dataplane) to Proposed Standard

On May 3, 2021 at 5:17:58 AM, Peter Psenak wrote:

> Technically I agree with you and if everybody agrees, I'm fine to
> enforce the presence of the Prefix Attribute Flags TLV in the Locator TLV.

So...what does everyone else think?

We need to close on this point before the IESG evaluates the document.  I'm requesting it to be put on the May/20 telechat, which means that we should have a resolution and updated draft by the end of next week.


Thanks!

Alvaro.



On May 3, 2021 at 5:17:58 AM, Peter Psenak ([e-mail]<mailto:[e-mail]>) wrote:
Hi Gunter,

Prefix Attribute Flags Sub-TLV has been defined as an optional Sub-TLV.
The problem you describe is not specific to Locator TLV, same applies to
regular IPv4/v6 prefixes (forget SR MPLS for a while) - if the Prefix
Attribute Flags TLV is not included, one can not tell whether the prefix
has been propagated (L1->L2) or generated as a result of the local
interface attached on the originator. Same applies to redistribution and
R-flag for IPv4 prefix TLVs.

SRv6 Locator TLV has been defined a while back and the Prefix Attribute
Flags Sub-TLV has always been an optional Sub-TLV of it. I'm not sure we
can start to mandate the Prefix Attribute Flags TLV at this point.

Technically I agree with you and if everybody agrees, I'm fine to
enforce the presence of the Prefix Attribute Flags TLV in the Locator TLV.

thanks,
Peter


On 03/05/2021 10:45, Van De Velde, Gunter (Nokia - BE/Antwerp) wrote:
> Hi Peter, All,
>
> Could we update to "draft-ietf-lsr-isis-srv6-extensions" that the prefix-attribute tlv is mandatory when a locator is redistributed?
>
> Why?
> *When calculating a LFA for an SRv6 End.SID we better know if the locator has been redistributed or not for a correct operation.
>
> Reasoning:
> * A locator has the D bit. This one is set when we redistribute from L2 to L1.
> ** So this end-sid will not be used as we know that it is redistributed.
>
> * In the other direction (L1-L2), we only know that a locator is redistributed from L1 to L2 if the prefix-attribute sub-tlv is advertised.
> ** This means if the operator does not configure advertisement of the prefix-attribute tlv, ISIS could potentially use an end-sid which does not terminate on the expected node.
>
> * Compared to sr-mpls, a prefix-sid has the R flag indicating it is redistributed.
> * We don't have that for locator end-sids.
>
> Relevant snip from " draft-ietf-lsr-isis-srv6-extensions"
>
> 7.1. SRv6 Locator TLV Format
>
> The SRv6 Locator TLV has the following format:
>
> 0 1 2 3
> 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Type | Length |R|R|R|R| MT ID |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>
> Type: 27
>
> Length: variable.
>
> R bits: reserved for future use. They MUST be
> set to zero on transmission and MUST be ignored on receipt.
>
> MT ID: Multitopology Identifier as defined in [RFC5120].
> Note that the value 0 is legal.
>
> Followed by one or more locator entries of the form:
>
> 0 1 2 3
> 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Metric |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Flags | Algorithm |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Loc Size | Locator (variable)...
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> | Sub-TLV-len | Sub-TLVs (variable) . . . |
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>
>
> Metric: 4 octets. As described in [RFC5305].
>
> Flags: 1 octet. The following flags are defined
>
> 0
> 0 1 2 3 4 5 6 7
> +-+-+-+-+-+-+-+-+
> |D| Reserved |
> +-+-+-+-+-+-+-+-+
>
> where:
> D-flag: Same as described in section 4.1. of [RFC5305].
>
>
> G/
>