[mrtg] MRTG vs. Cricket

[Steve Shipway]

Common problem (this really should be in an FAQ somewhere...?)

If you are polling at the default 5-min interval, a 32-bit counter will wrap at approximately 140Mbps.  Since by default MRTG will use SNMPv1 you end up using a 32bit counter unless you specify to use SNMPv2 and get the 64bit HC counters.

In your Target[] definitions, put a trailing :::::2 to specify to use SNMPv2, eg:

Target[foo]: 1:comm at device:::::2

This will use the 64bit counters, which can handle a rate of many many Gbps before wrapping, event at 5min intervals!

Another option is to use 1min intervals, but this only postpones the problem until about 700Mbps.

Always use SNMPv2 for any device supporting >100Mbps interfaces.

Steve

________________________________________
From: mrtg-bounces at lists.oetiker.ch [mrtg-bounces at lists.oetiker.ch] On Behalf Of Patrick Topping [ptopping at pobox.com]
Sent: Saturday, 31 January 2009 6:33 a.m.
To: 'mrtg at lists.oetiker.ch'
Subject: [mrtg] MRTG vs. Cricket

I am hoping someone can help me out with this issue I am having.....  I
am working on migrating from Cricket to MRTG.  So far so good except
that the Cricket graphs are pretty spot on for a 5 minute interval while
the MRTG graphs are anywhere from 600 Mbps to 1 Gbps off at peak time.
For example, I have a 10-gig circuit that runs roughly at 7 Gbps a peak
time.  Cricket graphs it damn near spot on.  MRTG is off by almost a
full gig.