Hi Yang,

There is lack of clarity with stat types as defined in 7854. It specifically 
called out Adj-RIB-In and defines non pre/post Adj-RIB-In stat types/metrics, 
such as Loc-RIB, duplicates, invalid by, etc.  This kind of renders the L flag 
misleading on some types considering some represent post processing metrics.  
8671 doesn’t attempt to redefine or correct 7854 in this sense.  Instead 8671 
calls out explicitly new types that should be used.  The O flag is specifically 
not set or ignored for stat reports, supporting backwards combability. Types 14 
– 17 if sent will be understood by the BMP receiver, as those are explicitly 
defined. L flag being set or not for types 14 – 17 is moot as the types clearly 
define the metric for the given peer defined by the per-peer header (sans L or 
O flags).

Thanks,
Tim

On 2/1/23, 6:23 PM, "GROW" <[email protected]> wrote:

Stats Reports follows per-peer header, which has L flag indicating
indicates that the message reflects post-policy

RFC7854 created stats type 7/8/9/10 for route counts, without
specifying whether they are pre or post policy

Later RFC8671 added stats report 14/15/16/17 with dedicated stat types
for pre / post policy in Adj-RIB-Out, making the route counter setup
inconsistent with Adj-RIB-In
https://www.iana.org/assignments/bmp-parameters/bmp-parameters.xhtml#statistics-types

Should L flag be meaningful for stats reports? If this is the case
data in type 7/8/9/10 would be expected to match L flag?
If not, why didn't stat type 7/8/9/10 get updated to match the setup
in RFC8671 for Adj-RIB-Out?

In JunOS 18's implementation, it appears stats reports are always sent
with L flag set to 0, even though only post-policy monitoring has been
configured.

a related discussion that predates RFC 8671
https://mailarchive.ietf.org/arch/msg/grow/Coqz738DPv7ZGVy-9GGzUBjtsF8/

Thanks.



Yang

_______________________________________________
GROW mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/grow
_______________________________________________
GROW mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/grow

Reply via email to