Re: [NGO] external module properties

["David Harrington" <ietfdbh at comcast.net>]

Hi,

My concern also includes multiple instances of a vlan module running
within a single (master/subagent) environment. This might be found in
a router chassis that supports new and legacy blades, or an x86 PC
server that can support disk drives from different vendors. 

Each of these components might support different revisions of the same
module, depending on the vendor model and release. In SNMP, this
wasn't critical beacause we were mostly monitoring. Th eoperators
complained about this for the CLI interface in the IAB workshop. 

If we are going to configure these using model-and-release-dependent
modules, we probably need to know which version of the "knob
definitions" are supported by the different components. If there are
includes in the module, then we better know which sub-module is being
referenced, just as installs on Linux usually require a clear
indication of the dependencies on specific versions of other packages.

dbh

> -----Original Message-----
> From: Andy Bierman [mailto:ietf at andybierman.com] 
> Sent: Tuesday, April 29, 2008 1:45 PM
> To: David Harrington
> Cc: 'Phil Shafer'; 'NETCONF Goes On'
> Subject: Re: [NGO] external module properties
> 
> David Harrington wrote:
> > I think that's a problem.
> > 
> > dbh 
> > 
> >> -----Original Message-----
> >> From: ngo-bounces at ietf.org [mailto:ngo-bounces at ietf.org] On 
> >> Behalf Of Phil Shafer
> >> Sent: Tuesday, April 29, 2008 12:13 PM
> >> To: Andy Bierman
> >> Cc: NETCONF Goes On
> >> Subject: Re: [NGO] external module properties
> >>
> >> Andy Bierman writes:
> >>> But if module FOO in any NETCONF system imports module X,
> >>> and module BAR imports module X, can we say in the YANG spec
> >>> that they MUST refer to the same exact module X?
> >> You can only say that if you can say that there is only
> >> exactly one module X.  If everyone makes their own "vlan"
> >> module, you're out of luck.
> >>
> 
> 
> I am proposing that within a given running NETCONF agent,
> that the [sub]module names are unique.  You cannot have
> multiple "vlan" modules within the same agent.  That
> implies that all import/include searches will use the
> same algorithm, but the spec can be silent about that.
> 
> (Multiple versions of the vlan module within a single agent
> is a different problem.)
> 
> >> Thanks,
> >>  Phil
> 
> Andy
> 
> 

_______________________________________________
NGO mailing list
NGO at ietf.org
https://www.ietf.org/mailman/listinfo/ngo