>>> At a minimum, we need to fix the fact that ospfv3RestartAge and 
>>> ospfv3NbrRestartHelperAge are defined as Ospfv3UpToRefreshIntervalTC which 
>>> is  Unsigned32 (1..1800). This poses a problem of what to return when these 
>>> MIB variables are not applicable since 0 is invalid.
>>> We may find some other additions and corrections now that we have some 
>>> implementation experience. Speaking as OSPF WG co-chair, we'd be happy if 
>>> you would be willing to take on this task.
>> 
>> If the object is inapplicable, why would it not be treated as if it didn't 
>> exist in the scenario? On a GET, return noSuchObject, and on a GET-NEXT or 
>> GET-BULK, skip it? Both objects are read-only. Do we have an erratum on 
>> these?
> 
> This is not an option since these are mandatory objects in the MIB 
> compliance. 

Mandatory to implement. I didn't say "don't implement them", I said "don't 
respond with them when they are inappropriate". By your argument, they can 
never be zero, because they will always have a value in 1..1800, because they 
are mandatory.

> I can poll the implementation that reported the problem. I suspect it was 
> just a matter of supplanting their MIB tools to allow 0 to be returned. 

"Just"…

If the objects can have a value that is not in 1..1800, their range must 
include the other value. We can create a new TC or simply use 0..1800, but we 
have to somehow explain what zero means. If we change the TC that the two 
objects use right now, that also affects a number of ReadCreate objects: under 
what circumstances may they be created with the value zero?
_______________________________________________
OSPF mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ospf

Reply via email to