Le 2012-07-23 06:47, Stephane Bortzmeyer a écrit :
Main comment: I am not convinced by the whole idea of "Discontinuity
Time". Isn't the whole idea of SNMP to have dumb agents, often without
stable memory, and the manager is supposed to know that, to detect
discontinuities and to deal with it? Why is DNS different from typical
MIB-II counters, which can go back to zero when the router reboots?
The main use case is when the DNS daemon restarts and loses state. This
creates a discontinuity in the DNS MIB counters but not in the global
MIB-II counters. That's why we introduced a separate discontinuity
indicator that the manager can monitor to identify discontinuities, just
like it does with MIB-II.
Other MIBs use the same technique (e.g. the OSPFv2 MIB [RFC4750]).
Second, the draft requires a mention of RFC 6168, which covers also
statistics.
Clearly we can add a reference to that RFC.
But it looks like it doesn't go into much detail about what statistics
are needed, just that some are. So, not much guidance for us. Or did I
overlook something?
Otherwise, the draft seems sensible, addresses a real problem for DNS
operators who have subcontractors, and, IMHO, correctly answer the
warnings of RFC 3197.
Thank you! That's very encouraging!
Simon
--
DTN made easy, lean, and smart --> http://postellation.viagenie.ca
NAT64/DNS64 open-source --> http://ecdysis.viagenie.ca
STUN/TURN server --> http://numb.viagenie.ca
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop
