Incorrect data point

[cacti2008]

I have notice a big different between the graph that CACTI produced vs MRTG, specifically relating to input/output above 100 Mbps. For example, my MRTG is showing a current output of 836 Mbps, but CACTI shows an output of 109M. I check my Cisco 6513 and it is running around 800 Mbps on that particular interface. I can understand a little difference in the numbers because of the polling time. But not that big of a different.

I am currently using SNMP version 2.

Any idea how to fix this?

Thanks in advance.

[gandalf]

Is it possible, that one graphs shows "bytes per second", the other one "bits per second"?
Reinhard

[cacti2008]

Both are in bits per second.

Attached is the graphs from CACTI and MRTG.

[gandalf]

Both are in bits per second.

Attached is the graphs from CACTI and MRTG.

???
Reinhard

[cacti2008]

See attached graphs.

[gandalf]

I worried, because not only the magnitude of the graphs differ, but the shape as well. Please double check, that the OID used is the same in both examples (don't flame me for this request, but it's really weird)
Reinhard

[cacti2008]

Maybe it is the host template that I have selected when creating the device. I am currently using "Cisco Router". Should I go with something else, like Generic SNMP-enabled Host or ucd/net SNMP Host?

Is there a better template (maybe something I have to import because it didn't come with the standard Version 0.8.7b install? My entire network is Cisco.

Thanks again.

[gandalf]

The Host templates may differ. But the Traffic Templates used are always the same (because the interface MIB is a RFC standard for ALL devices). This is NOT the root cause of your problem
Reinhard

[cacti2008]

How do I get to the OID in CACTI?

[cacti2008]

I am sure it is CACTI having problem with getting or calculating the data above 100 Mbps. I just found several more interfaces that have been graphing relatively close (MRTG and CACTI running on different server) until the interface traffic went above 100 Mbps. I have an interface running at 381.3 Mbps (MRTG), but only reported as 109.67 Mbps (CACTI).

Here are some sample graphs for an interface that (examples1) went above 100 Mbps and (examples2) consistently running below 100 Mbps on the same Cisco 6513. As you can see, graphs are basically the same until the traffic shot above 100 Mbps. Doesn't look like CACTI is calculating something correctly because it is messing up the graph above 100 Mbps.

[gandalf]

How do I get to the OID in CACTI?

See "System Utilities -> View Poller Cache" to see the commands that cacti uses
Reinhard

[gandalf]

I am sure it is CACTI having problem with getting or calculating the data above 100 Mbps. I just found several more interfaces that have been graphing relatively close (MRTG and CACTI running on different server) until the interface traffic went above 100 Mbps. I have an interface running at 381.3 Mbps (MRTG), but only reported as 109.67 Mbps (CACTI).

Here are some sample graphs for an interface that (examples1) went above 100 Mbps and (examples2) consistently running below 100 Mbps on the same Cisco 6513. As you can see, graphs are basically the same until the traffic shot above 100 Mbps. Doesn't look like CACTI is calculating something correctly because it is messing up the graph above 100 Mbps.

Each time I here from "above 100 MBps" problems, I make sure, that 64 bits counters are used in this case.
When using standard 3oo sec interval, 32 bit counters roll over at 114 Mbps. There's a sticky entry reflecting this and it shows how to remedy
Reinhard

[cacti2008]

Thank you everyone. I've tried the 64-bits counter and it worked perfectly.