Hi,
"Bogaert, Bart (Nokia - BE)" <[email protected]> wrote:
> Hi,
>
>
>
> We are wondering why in the no provision has been forseen to link the
> "interface-world" to the "entity-world": when looking at the YANG modues for
> entity and interface it is not there. In the SNMP entity MIB there seems to
> be an object that serves this purpose: entAliasMappingEntry. Is there any
> reason why this was not considered in the YANG modules?
No real reason; we decided to get a basic structure in place and
then wait for comments from the WG. But when we published -00 the WG
was still busy with lots of other work, and didn't have the cycles to
adopt this draft contine to work on it. Thus the work stalled.
But now it seems this document might get adopted by the WG soon, so
hopefully we can re-start this work.
> It is also possible to do the linking from /interfaces to /entity as this
> would allow proper validation of interface settings immediately. If the
> linking is from /entity to /interfaces then both can be configured (and
> validated) separately but once an interface has to be associated with a real
> HW resource then again (maybe only partial) validation is required to ensure
> that this interface can actually be associated with the HW entity
> instantiation. If the linking is from interfaces to entity then all this
> validation can be done in 1 step (which to me seems more logical).
I can see how a config false reference from interfaces-state to
entity-state could work. It is common to have implicit hw info by
embedding the hw location in the interface name ("ethernet-1/0").
The entAliasMappingEntry is more generic, and the same concept could
be used in YANG as well (with an instance-identifier instead of the
OID).
/martin
_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod
