Re: Interface configuration

[Bob Stewart <[e-mail]>]

>No, don't delete those objects.  I agree with you (I've just been
>so caught up in other work, I haven't been able to monitor this group
>-- sorry :^).
>
>It's essential that there be some way to determine exactly how the 
>different entities within the chassis are connected.

The question is not whether the information is essential, rather what form is
appropriate and whether it is available to the chassis agent.  For example, I
don't believe we want to force the chassis agent to share a network connection
with all the entities and query their implementation-specific MIBs with SNMP.

>It's also very 
>important that this information be available from a standard MIB.
>Since there is no way to get that information from one of the 
>existing standard MIBs (as far as I know??), this is the obvious place 
>to put it.

I agree that a standard MIB is preferable, but not necessarily the one found
in the chassis agent.  Perhaps the answer is to define a table that defines
segment mapping information WITHIN the entity itself.  A chassis agent
supports the Chassis MIB summary groups.  Every chassis-resident entity
supports the Chassis MIB self-description groups but not necessarily the
summary groups.

My suspicion is that an attempt to define the self-description groups would
run afoul of major implementation differences and complaints about the burden
on every system.  Collecting the information in the summary groups obscures
those problems, it doesn't eliminate them.  In any case, I have to be able to
control segment mapping by talking to the entity.  Perhaps we should turn our
attention there for a standard.

	Bob