[mrtg] MRTG-RRD anomaly!!

Steve Shipway s.shipway at auckland.ac.nz
Tue Sep 30 01:21:54 CEST 2008


As Azrol said, this is due to using SNMPv1, and the MRTG counter rollover detection.

If you use SNMPv1 you get 32bit counters, which roll over at about 120Mbps.  So, above 100Mbps, you should use SNMPv2 64bit counters.
If you do a calculation in the Target[] definition, then rollover is calculated AFTER the calculation, and so unless you are using 64bit counters, the TOTAL must be less than 120Mbps, which yours is not - so MRTG incorrectly assumes counter rollover and gets it wrong.

Use SNMPv2, or if you use Routers2 use a userdefined summary instead, then it will work as you will avoid MRTG rollover problems.

To use SNMPv2, add ':::::2' to the end of each Target component (provided your router supports SNMPv2, of course)

Steve

________________________________
It may cause by snmp version. Please use snmpv2 instead of snmpv1.
-- Azrol --

MRTG of two interfaces are working well but the aggregation graph sometimes is showing wrong graph. It seems it is just forgetting to sum up the graphs.

i.e.
interface 1  current in: 64.93Mb   out:18.04Mb
interface 2 current in:59.33 Mb  out:3.12Mb
int1+int2  current in :11.32 Mb    out:  21.17Mb


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/mrtg/attachments/20080930/efa80076/attachment.html 


More information about the mrtg mailing list