Re: New overview text in Chassis MIB

[[e-mail] (Keith McCloghrie)]

Chuck,
 
> I understand and agree with 5.5 below.

Great.

> I don't understand what 5.6 below means. Does it mean that, in a
> particular chassis, I may have multiple agents that export the same
> instances of chassis MIB objects?  

Yes.  The same instances having the same values.

> How/why would one implement this
> arrangement? Is it viable/useful for a chassis as big as Texas?
 
Well, it's kinda hard for me to answer because my mental picture has a
single "chassis management agent".  However, several folks on the 
mailing-list seemed to want to have multiple agents in one chassis 
implement this MIB.  Thus:

1. since questions have been raised on the mailing-list, it seems 
important that the text is not silent on this issue.

2. providing they meet the requirements listed in the proposed 5.6 
(i.e., each agent contains the same set of variables with the same 
values for all mandatory parts of the MIB), I can't see any reason to 
prohibit multiple such agents.

As to why one would want multiple agents, they only reason I recall 
being mentioned on the mailing-list was for redundancy.  This might
in fact be more important if the chassis were as big as Texas, e.g.,
a "partitioned" chassis (!!).  Should the consistency rule be defined 
to be a little looser in such a situation ?

> If I am a management station, is this replication of the chassis
> information at multiple agents helpful? Confusing? How would I sort it
> out?

Isn't it the same situation as handling today's scenario of a router 
which accepts SNMP requests sent to any of its multiple IP addresses ?

Keith.