[mrtg] MRTG problems on a Foundry Switch?

[Alex van den Bogaerdt]

On Mon, Oct 08, 2007 at 04:58:19PM -0700, Scott Baker wrote:
> I've been running MRTG for years and never had a problem like this.
> I just introduced a new switch into our network and wanted to
> monitor it.
> 
> http://www.web-ster.com/bakers/foundry_1-day.png
> 
> It seems like any time this port goes over 110 or so Mb/s the graph
> freaks out. It's like the counters are looping because it's so much
> bandwidth. The config is super simple:
> 
> Target[foundry_1]: 1:xxxxxx at xxxxx:
> SetEnv[foundry_1]: MRTG_INT_IP="" MRTG_INT_DESCR="GigabitEthernet1"
> MaxBytes[foundry_1]: 125000000
> Title[foundry_1]: Traffic Analysis for 1 -- FESX424 Router
> PageTop[foundry_1]: <H1>Traffic Analysis for 1</H1>
> Options[foundry_1]: bits
> 
> I'm thinking maybe this is a problem with the hardware and not MRTG?

It is not the hardware, it is not MRTG, it is SNMP.  32 bit counters
wrap too often for your bandwidth:

2^32 Bytes / 300 seconds = 14,316,557 bytes per second. That's
114,532,456 bits per second, slightly more than 110 MB/s.

You see 1,1 instead of 1,1+2^32.

> Anyone had this problem before?

<g> look at the archives and RTFM ...