Re: [netmod] *one* week 2nd WG Last Call: draft-ietf-netmod-revised-datastores-07

[Martin Bjorklund <[e-mail]>]

Ladislav Lhotka <[e-mail]> wrote:
> On Thu, 2017-12-21 at 14:25 +0100, Martin Bjorklund wrote:
> > Ladislav Lhotka <[e-mail]> wrote:
> > > Martin Bjorklund <[e-mail]> writes:
> > > 
> > > > Hi Andy,
> > > >
> > > > Andy Bierman <[e-mail]> wrote:
> > > >> Hi,
> > > >> 
> > > >> I have reviewed draft-07 and my previous comments about NMDA have been
> > > >> addressed.
> > > >> 
> > > >> This might be the most important sentence in the draft:
> > > >> 
> > > >> sec. 5.3
> > > >> 
> > > >>    The datastore schema for <operational> MUST be a superset of the
> > > >>    combined datastore schema used in all configuration datastores except
> > > >>    that YANG nodes supported in a configuration datastore MAY be omitted
> > > >>    from <operational> if a server is not able to accurately report them.
> > > >> 
> > > >> The MUST implies that there is no need to design a YANG library that can
> > > >> support
> > > >> an implementation that violates this MUST (i.e., 1 schema tree for the
> > > >> super-set)
> > > >> 
> > > >> The MAY is troublesome because it completely contradicts the conformance
> > > >> expressed
> > > >> in each YANG module supported by the server.  Any data node without any
> > > >> if-feature-stmts is mandatory to implement.
> > > >
> > > > This is required for transition purposes; a server that wants to
> > > > implement <operational> should not have to implement all modules at
> > > > once (as applied config).
> > > >
> > > >> What about config=false subtrees within a config=true subtree?
> > > >> Can they be omitted from <operational> as well, or does the draft just
> > > >> intend to
> > > >> omit the operational value of config=true nodes?  Should be specific.
> > > >
> > > > The text says "nodes supported in a configuration datastore MAY be
> > > > omitted from <operational>".  So it is implicit that it only applies
> > > > to config true nodes (since config false cannot be supported in a
> > > > config ds).  How about:
> > > >
> > > > OLD:
> > > >
> > > >     The datastore schema for <operational> MUST be a superset of the
> > > >     combined datastore schema used in all configuration datastores except
> > > >     that YANG nodes supported in a configuration datastore MAY be omitted
> > > >     from <operational> if a server is not able to accurately report them.
> > > >
> > > > NEW:
> > > >
> > > >     The datastore schema for <operational> MUST be a superset of the
> > > >     combined datastore schema used in all configuration datastores
> > > >     except that YANG "config true" nodes supported in a configuration
> > > 
> > > If this is about schema or data nodes, I suggest to state it
> > > explicitly:
> > > 
> > >     ... "config true" schema/data nodes ...
> > 
> > Yes, the new text uses "configuration data nodes".
> > 
> > > >     datastore MAY be omitted from <operational> if a server is not
> > > >     able to accurately report them.
> > > >
> > > >
> > > >> Perhaps this draft does not need the MAY half of the sentence at all.
> > > >> The YANG library can specify that it is for conformance-reporting, not
> > > >> conformance-defining.
> > > >
> > > > I think we should keep the MAY, since the YANG library has to be
> > > > designed to support this case.
> > > 
> > > Shouldn't the server add corresponding deviations to the schema for
> > > <operational> in this case?
> > 
> > We wanted to explicitly support the case that a server doesn't (yet)
> > implement a given module with config nodes in operational.  But maybe
> 
> But with the new schema of YANG library, say Alt B, such a server can simply
> omit this module from the schema of <operational>, right?

Right; note that this document specifies the requirements, and yang
library is designed accordingly.


/martin


> 
> Lada
> 
> > we should design for the future and remove the MAY half of the
> > sentence, as suggested above, and let such servers use deviations in
> > this case.
> > 
> > 
> > /martin
> > 
> > 
> > 
> > 
> > > 
> > > Lada
> > > 
> > > >
> > > >
> > > > /martin
> > > >
> > > >
> > > >
> > > >
> > > >> 
> > > >> 
> > > >> 
> > > >> 
> > > >> Andy
> > > >> 
> > > >> 
> > > >> On Mon, Dec 4, 2017 at 6:35 AM, Lou Berger <[e-mail]> wrote:
> > > >> 
> > > >> > All,
> > > >> >
> > > >> > This starts a second working group last call on
> > > >> > draft-ietf-netmod-revised-datastores.
> > > >> >
> > > >> > As this is a 2nd LC that is focused on changes since the last LC, it
> > > >> > closes in *one* week. The working group last call ends on December 11.
> > > >> > Please send your comments to the netmod mailing list.
> > > >> >
> > > >> > At this point, we're most interested in verifying that previous
> > comments
> > > >> > are addressed since the last call on the -04 rev of the draft was held.
> > > >> >
> > > >> > A summary of changes can be found at
> > > >> > https://mailarchive.ietf.org/arch/msg/netmod/DWtD12bGkBZabEygRfiwZfcnUU
> > 4
> > > >> >
> > > >> > A diff can be found at
> > > >> > https://tools.ietf.org/rfcdiff?difftype=--hwdiff&url1=draft-ietf-netmod
> > -
> > > >> > revised-datastores-04.txt&url2=draft-ietf-netmod-revised-datastores-
> > 07.txt
> > > >> >
> > > >> > Comments along the of: I have reviewed this version of the document and
> > it
> > > >> > addresses my previous comments would be particularly helpful.
> > > >> >
> > > >> > Thank you,
> > > >> > Netmod Chairs
> > > >> >
> > > >> > _______________________________________________
> > > >> > netmod mailing list
> > > >> > [e-mail]
> > > >> > https://www.ietf.org/mailman/listinfo/netmod
> > > >> >
> > > >
> > > > _______________________________________________
> > > > netmod mailing list
> > > > [e-mail]
> > > > https://www.ietf.org/mailman/listinfo/netmod
> > > 
> > > -- 
> > > Ladislav Lhotka
> > > Head, CZ.NIC Labs
> > > PGP Key ID: 0xB8F92B08A9F76C67
> > > 
> -- 
> Ladislav Lhotka
> Head, CZ.NIC Labs
> PGP Key ID: 0xB8F92B08A9F76C67
>