problems with high traffic graphs

[atolibot]

Hello,

I've searched over the forums looking for someone with a similar problem, but so far havent had any luck. I was hoping someone here would be able to point me in the right direction.

We've been using cacti for sometime now, but have recently run into a problem. Our traffic on a few ports has started to creep past 110-120mbit/sec range and we've started to see the outbound traffic graphing "crash" up and down when it gets past that threshhold. We've confirmed what the graphs are showing is not what is happening on the switch/router side, and experimented with this on several different layer 3 and cisco catalysts, cacti displays similar results on all past the 110-120mbit/sec mark. We are using the In/Out 95th percentile template for the graphs with no adjustments.

Im no snmp guru so excuse myself if Im missing something obvious. Ive attached one of the graphs displaying this behavior. Any insight would be most appreciated!

Thanks in advance.

[Marcelo]

Try using snmp 64 bits counters, that should fix the problem.
Greetings.

[pestilence]

either 64bit counter or your rrd files have a maximum lower than the provided value (do a rrdtool info filename and check the ds max value).