[Xymon] Weird value in devmon graph

Olivier AUDRY olivier at audry.fr
Mon May 9 14:31:37 CEST 2011


many thx I will do this 

oau

----- Mail original -----
De: "W.J.M. Nelis" <Wim.Nelis at nlr.nl>
À: "Olivier AUDRY" <olivier at audry.fr>
Envoyé: Lundi 9 Mai 2011 13:52:50
Objet: Re: [Xymon] Weird value in devmon graph

Hello,

> during two days this week end we got very huge value in all my devmon graph. See the attachement. Since this morning no more issue.
>
> any idee ? Perhaps we can add a max value at the rrd creation ?
>   
We see similar graphs from a few devices which at random times reset all 
their SNMP counters. If the throughput is measured using a 32-bit 
counter, and if you are using type COUNTER in the RRD, a reset of a 
counter is interpreted as a counter-wrap. With an polling frequency of 
once per 120 seconds, you will see peaks of up to 290 [Mb/s].  This 
explains that there might be peaks, it does not explain that the 
peak-values are effectivly the same in the instances shown.

If a reset of the SNMP counters turns out to be the cause, you could 
solve it (A) by using 64-bit counters or (B) by using type DERIVE in RRD 
combined with a lower limit of zero. The limit will "annihilate" any 
negative derivates.

Kind regards,
  Wim Nelis.



*******************************************************************************************************
The NLR disclaimer (http://www.nlr.nl/emaildisclaimer) is valid for NLR e-mail messages.
*******************************************************************************************************




More information about the Xymon mailing list