Lada: 

As dynamic control plane datastores start-up, these will need to go from an
initial state to a "current state".   I2RS ephemeral dynamic control plane
datastores will have a model that comes up in a limited mode with the I2RS
agent's configuration.  The rest of the configuration will be update from a
client over the network.  You are correct that the RFC6241 definition does
not cover this state.  The I2RS concept does cover "Config=TRUE" being the
test of configuration.   I think this is the same test that applies because
configuration data can be updated even in the configuration datastore
(intended to applied). 

Good catch on the definition. 

Sue 

-----Original Message-----
From: netmod [mailto:[email protected]] On Behalf Of Ladislav Lhotka
Sent: Tuesday, March 21, 2017 5:59 AM
To: Jürgen Schönwälder
Cc: [email protected]
Subject: Re: [netmod] draft netmod charter update proposal


> On 21 Mar 2017, at 10:43, Juergen Schoenwaelder
<[email protected]> wrote:
> 
> On Tue, Mar 21, 2017 at 10:13:40AM +0100, Ladislav Lhotka wrote:
>> 
>> The revised-datastores draft changes the semantics of "configuration
data" - for example, the definition from RFC 6241 clearly won't apply to the
"running" datastore in the new datastore model.
> 
> Why would that be the case?

RFC 6241:

   o  configuration data: The set of writable data that is required to
      transform a system from its initial default state into its current
      state.

If I understand revised-datastores correctly, this will no more be the case
for "running" because it may contain intermediate data that require further
processing (templates) and inactive data that are not used for changing the
device state.

This may look like nit-picking but the fact is that we often use the term
configuration in the sense "you know it when you see it". 

> 
>> So a new definition of configuration data will probably be needed, and
this implicitly changes the semantics of the "config" statement.
>> 
> 
> YANG defines the config statement as follows:
> 
>   The "config" statement takes as an argument the string "true" or
>   "false".  If "config" is "true", the definition represents
>   configuration.  Data nodes representing configuration are part of
>   configuration datastores.

If the "config" statement really carried some protocol-specific semantics
that isn't meaningful for all potential uses of YANG, it would be better to
remove it from core YANG and define it as an extension that would be
mandatory for configuration protocols that need it.

Lada

> 
> I do not think it is the intend of the revised datastore model as 
> written down in the I-D to change this.
> 
>> BTW, we use rw/ro in tree diagrams.
> 
> Which is a mis-nomer (tree diagrams were inherited from the SNMP world 
> and somehow the rw/ro distinction was kept even though it is 
> technically wrong). There are more details here, I will start a 
> separate thread for this.
> 
> Note that the diagrams in the revised datastore ID make a clear 
> distinction between ct/cf and rw/ro. In particular, the ID notes that 
> ct object may be rw in one datastore but ro in a different datastore.
> 
> /js
> 
> -- 
> Juergen Schoenwaelder           Jacobs University Bremen gGmbH
> Phone: +49 421 200 3587         Campus Ring 1 | 28759 Bremen | Germany
> Fax:   +49 421 200 3103         <http://www.jacobs-university.de/>

--
Ladislav Lhotka, CZ.NIC Labs
PGP Key ID: 0xB8F92B08A9F76C67





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

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

Reply via email to