Re: [Netconf] FW: New Version Notification for draft-ietf-netconf-monitoring-05

[Balazs Lengyel <balazs.lengyel at ericsson.com>]

Hello,
See below!
Balazs

Martin Bjorklund wrote:
> > General)
> > I think we should not exclude the possibility of monitoring non-netconf
> > sessions in this schema.
>
> I don't think we do.  The intention was that *if* the NETCONF server
> can manage other sessions, it MAY include them in the session list.  

BALAZS: The sentence
"Includes session specific data for NETCONF management session."
at the start of 2.1.4. seems to indicate that CLI sessions are strictly excluded.
While you can understand this to include CLI sessions, if they are visible to
the NETCONF server, this is far from trivial.

> > Ch 2.1.3)
> > YIN should be added as a format.
>
> Originally we had text that said that the format 'yang' also covered
> the yin syntax form.  So either we add that back, or we add 'yin'.
> Should we also add 'rnc' (in addition to 'rng')?

BALAZS: IMO add rnc as well

> > 2.1.4) in several places you use "a <rpc> message was expected"
> > Who expects it? Does the device know that an <edit-config> should be arriving
> > at 7:00 PM? :-) Rather remove the expected part.
>
> The server waits for (and thus expects) <rpc> messages.  If it gets
> something else, the counter is incremented.

BALAZS: I still don't like the wording, but don't care.

> > 2.1.5) Please indicate if droppedSessions includes session closed due to
> > parseErrors.
>
> There is nothing in 4741 that says that a parse error MUST terminate the
> session.  

BALAZS: Maybe there should be a mention of this in the 4741bis. We have talked about the 
situation a lot when discussing the ]]>]]> sequence and security consideration.

> > - get-schema output: if we have a YANG schema will it will be a string not any
> >   XML
>
> A string can be sent in anyxml.

BALAZS: Why do you want to force non-XML descriptions like yang or rnc into an XML container? 
It can be done, but its kind of strange.
What would be the starting XML element? That should be defined.