Re: [Idr] 5575bis nit

Hi Jeff,

I opened 2 issues for tracking that (also including your thoughts from the second mail) on github:

https://github.com/stoffi92/rfc5575bis/issues/160 <https://github.com/stoffi92/rfc5575bis/issues/160>

https://github.com/stoffi92/rfc5575bis/issues/161 <https://github.com/stoffi92/rfc5575bis/issues/161>

I came across the second issue a while back when I was writing import policies and saw what juniper does (thanks for sharing the details), but I haven’t checked what other implementations actually put there.

Cheers

Christoph

-- 
Christoph Loibl
c@tix.at | CL8-RIPE | PGP-Key-ID: 0x4B2C0055 | http://www.nextlayer.at

> On 13.02.2020, at 00:41, Jeffrey Haas <jhaas@pfrc.org> wrote:
> 
> In RFC 5575 and 5575-bis, we have several extended communities defined.
> 
> :    +-----------+----------------------+--------------------------------+
> :    | community | action               | encoding                       |
> :    +-----------+----------------------+--------------------------------+
> :    | 0x8006    | traffic-rate-bytes   | 2-byte ASN, 4-byte float       |
> :    | TBD       | traffic-rate-packets | 2-byte ASN, 4-byte float       |
> :    | 0x8007    | traffic-action       | bitmask                        |
> :    | 0x8008    | rt-redirect AS-2byte | 2-octet AS, 4-octet value      |
> :    | 0x8108    | rt-redirect IPv4     | 4-octet IPv4 addres, 2-octet   |
> :    |           |                      | value                          |
> :    | 0x8208    | rt-redirect AS-4byte | 4-octet AS, 2-octet value      |
> :    | 0x8009    | traffic-marking      | DSCP value                     |
> :    +-----------+----------------------+--------------------------------+
> : 
> :                Table 2: Traffic Action Extended Communities
> 
> Nits: We use byte in some spots for AS, octet in others.  "addres" is a
> typo.
> 
> A question came up from a customer, else I would have let this question
> lay idle:  When you're running flowspec within a BGP Confederation (RFC 5065),
> whose AS do you stick in there?  The Member-AS, or the AS Confederation
> Identifier?
> 
> I could make a few weak arguments that the Member-AS makes sense. (And, is
> in fact, what Juniper's code currently does.)  However, I think the general
> intent is such that consistent action within the Confederation is what is
> desired, similar to what it'd be like if the behavior is iBGP.
> 
> While the AS number is largely an advisory number for the features in
> question, it does potentially have impact on policy.
> 
> My recommendation would be we add some text roughly like the following:
> Add "In the circumstance that this specification is utilized in a BGP
> Confederation [RFC 5065], AS Numbers used by these Traffic Action Extended
> Communities SHOULD be the BGP speaker's Confederation Identifier AS."
> 
> -- Jeff
> 
> _______________________________________________
> Idr mailing list
> Idr@ietf.org
> https://www.ietf.org/mailman/listinfo/idr

Re: [Idr] 5575bis nit - ASes in extended communities