It's been my experience that if you exceed the rate of a 32-bit counter, it rolls over, usually resulting in what appears to be a sharp decline on the graph, but on a 64-bit counter based graph it looks correct.
On Wed, 2011-06-08 at 12:04 +0100, Phil Mayers wrote: > On 06/08/2011 11:35 AM, Nick Hilliard wrote: > > On 08/06/2011 10:15, Phil Mayers wrote: > >> The first will never read >100mbit/sec. The second will read > >> correctly, so > >> if you're doing >100mbit/sec they'll look different. > > > > about 112mbit/sec. So if you see graphs pegged at ~112mbit/sec, you know > > you're running into a 32-bit counter limitation. > > > > snmpv1+32 bit counters are pretty useless these days. I wish Cacti would > > print up a large warning when people attempted to use these oids because > > they cause lots of breakage. > > Indeed. Frankly, it's damn annoying Cacti doesn't just "figure it out", > at least for standard counters where the 64-bit alternatives are well > known. But that's really the least of it's warts! > _______________________________________________ > cisco-nsp mailing list [email protected] > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ -- Walter Keen Network Engineer Rainier Connect Office: 360-832-4024 Cell: 253-302-0194 _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
