Re: I-D ACTION:draft-kulkarni-netconf-subagent-prot-00.txt

[Martin Bjorklund <mbj@tail-f.com>]

Andy Bierman <ietf@andybierman.com> wrote:
> I think it would be almost as much work to write this document
> as it was to write the netconf-prot document.  In practice,
> the capabilities and operations will not be the same across
> all sub-agents.  All the bells and whistles (error-option,
> confirmed-commit, etc.), for all the operations seems like
> a lot of complexity to me.

I agree.

> >> And what about special RPCs like <reset-interfaces>
> >> which might be implemented across multiple sub-agents?
> > 
> > This might be another reason for using a standard <exec> method (where
> > the operation itself is defined in the data model) as been discussed
> > earlier.
> 
> My point was that if every RPC is to be supported,
> then a generalized algorithm would be needed to
> define how it is properly invoked in the presence
> of multiple sub-agents.  IMO, this is difficult.

I agree.

> It is hard enough to define the algorithm for
> RPC methods we know about (like <edit-config>),
> but what about any arbitrary RPC?  I don't
> think <exec> really helps here, unless the contents
> of the exec command are tightly constrained.

The idea with exec is that the operation is defined for an instance in
the data model.  Something like
  <exec>
    <operation>restart</operation>
    <instance>/interface[ifIndex='1']</instance>
  </exec>

If the master knows which subagent handles /interface, he can dispatch
to that subagent.

Anyway, I agree that this isn't high on the priority list, and we
should probably collect real-world experience before doing this.


/martin

--
to unsubscribe send a message to netconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/netconf/>