Re: SigMIB Draft 10

Hi -

> From: "Freyman Phillip-FPF300" <[email protected]>
> To: "Woundy, Richard" <[email protected]>; <[email protected]>
> Sent: Thursday, March 30, 2006 9:08 PM
> Subject: RE: [ipcdn] SigMIB Draft 10
...
> As Gordon recalled, "not used" meant the value is not applicable to a
> particular object. Somebody wanted to know what happens if a get/set was
> used, so the suggestion was to return inconsistent value. 
...

Returning a "inconsistent value" in response to an SNMP get(next/bulk)
operation is simply wrong.  See RFC 3416 sections 4.2.1, 4.2.2,
and 4.2.3 respectively for the permitted responses.  So the question
still remains: how do these objects behave in the "not used" case?
The behaviour I'd expect would be for these objects to be present with
whatever value they had most recently been set to, which might be the
result of DEFVAL initialization.

In the case of a set request, "inconsistent value" would be a plausible
response, but it does mean that systems using SNMP to configure a
device will need to send the varbinds in separate PDUs to ensure consistent
results.  I don't see how this benefits anyone.

Randy