Re: [pim] draft-ietf-pim-yang-03: questions

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

All, 

I spoke to Reshad Rahman (BFD YANG Model editor) regarding the BFD model
yesterday and he assured me that the current model with the BFD session
parameters centrally (as opposed to within the protocols) will be
maintained. Consequently, I think OSPF can move to the IS-IS approach of a
single BFD enabled interface leaf node conditioned with a BFD feature.


Thanks,
Acee


On 1/13/17, 1:30 PM, "Derek Yeung" <[e-mail]> wrote:

>OSPF model use augmentation module because of two reasons.
>1) When OSPF first add BFD feature, the BFD model is not stable and we
>want to isolate the OSPF base model from it.
>2) It the earlier BFD model, the per protocol configuration is supported
>and there is a BFD grouping (bfd-grouping-base-cfg-parms) that OSPF need
>to use. See draft-zheng-bfd-yang-04.txt.
>
>The per-protocol configuration is later removed from BFD and hence from
>OSPF BFD model, leaving it with a single boolean leaf.
>If it is confident that the BFD model is now stable, we could change OSPF
>model to use feature with single boolean leaf instead.
>
>Thanks,
>Derek
>
>On 1/4/17, 9:35 AM, "Xufeng Liu" <[e-mail]> wrote:
>
>    During IETF at Seoul, this topic was brought to the attention of BFD
>WG Chair, Jeff Haas ( cc'ed ). I think that BFD WG is currently working
>on it, and will provide guidelines.
>    
>    Thanks,
>    
>    - Xufeng
>    
>    > -----Original Message-----
>    > From: Jeff Tantsura [mailto:[e-mail]]
>    > Sent: Tuesday, January 3, 2017 8:23 PM
>    > To: Stig Venaas <[e-mail]>; Toerless Eckert
><[e-mail]>;
>    > Acee Lindem (acee) <[e-mail]>; [e-mail]
>    > Cc: [e-mail]; [e-mail]
>    > Subject: Re: [pim] draft-ietf-pim-yang-03: questions
>    > 
>    > Hi,
>    > 
>    > I¹d like to align at least IGP¹s and PIM,
>    > 
>    > ISIS ­ bfd is a feature
>    > OSPF - defines module ietf-ospf-bfd
>    > 
>    > bfd containers:
>    > ospf:
>    > container bfd {
>    >          description "BFD configuration.";
>    >          leaf enable {
>    >            type boolean;
>    >            default false;
>    > 
>    > isis:
>    > container bfd {
>    >           if-feature bfd;
>    >           leaf enable {
>    >               type boolean;
>    >               default false;
>    > 
>    >  including Acee and Derek
>    > 
>    > Cheers,
>    > Jeff
>    > 
>    > 
>    > On 1/3/17, 13:32, "pim on behalf of Stig Venaas"
><[e-mail] on
>    > behalf of [e-mail]> wrote:
>    > 
>    >     Hi
>    > 
>    >     I agree we need to figure these things out. pim is not in a
>unique
>    >     position here, YANG models for other routing protocols must be
>facing
>    >     the same questions. The bfd wg is working a model too, right?
>Perhaps
>    >     discuss it in the bfd wg and work with those working on the
>model
>    >     there?
>    > 
>    >     Alternatively we could remove bfd from our models for now. They
>could
>    >     be added by a later RFC if a standard is needed.
>    > 
>    >     Stig
>    > 
>    > 
>    >     On Sun, Nov 13, 2016 at 6:16 PM, Toerless Eckert
><[e-mail]> wrote:
>    >     > To follow up on the questions i had on the mike on Seoul@PIM:
>    >     >
>    >     > Can you pls. point me to a place that defines how BFD has to
>work if
>    >     > two different protocols on the same interface are configured
>for differrnt
>    >     > local-multiplier, desired-min-tx, required-min-rx-interval
>and the like.
>    >     >
>    >     > I hope thats specified in some BFD RFC/draft, but i couldn't
>find it browsing
>    >     > them.
>    >     >
>    >     > Also: In Seoul, it was said that some vendors have
>configuration of these
>    >     > parameters on a per-protocol basis. WOuld love a pointer
>toward any single
>    >     > vendor doc about that.
>    >     >
>    >     > If these parameters can be configured underneath a particular
>protocol
>    >     > (like PIM/BFD), why then should these parameters be "ro"
>instead of "rw" ?
>    >     >
>    >     > Last but not least: Given how there is no PIM-BFD specific
>"protocol spec"
>    > out
>    >     > of PIM or MBoned, it looks to me as if all this discussion is
>probavbly better
>    >     > done in BFD working group, and they should suggest the
>parameters that
>    >     > can be configured.
>    >     >
>    >     > draft-ietf-ospf-yang has a section about BFD which seems to
>primarily
>    >     > care about making OSPF-BFD in the OSPF-YANG model optional. I
>think
>    >     > it would be prudent to do the same for PIM-BFD. On the other
>hand, even
>    > the
>    >     > OSPF-BFD definition does not seem to define these parameters.
>So maybe
>    >     > lets try to figur eout the best of what OSPF-BFD will
>ultimately do
>    >     > and then copy to PIM - unless we figure out reasons to go
>beyond that
>    > approach.
>    >     >
>    >     > Cheers
>    >     >     Toerless
>    >     >
>    >     > _______________________________________________
>    >     > pim mailing list
>    >     > [e-mail]
>    >     > https://www.ietf.org/mailman/listinfo/pim
>    > 
>    >     _______________________________________________
>    >     pim mailing list
>    >     [e-mail]
>    >     https://www.ietf.org/mailman/listinfo/pim
>    > 
>    > 
>    
>    
>