> Unless I misconfigured or misunderstand the gauge option, the graph should
> display the given value as-is, immediately.
...
> Why is the HTML graph not showing Max=100, avg=100, cur=100?

Because, you misunderstand the gauge option.

What you have not taken into consideration is the dreaded DATA NORMALISATION 
done by MRTG - this is adjusting the received value to what it would have been 
on the interval boundary, assuming a linear pattern.

EG assume you have a 5min interval.
At 12:00 value 10 is logged.  This goes into the 12:00:00 - 12:04:59 bucket 
as-is, since it is on the first point of the interval.
At 12:06 value 22 is logged.  Since this is a bit late, this is averages over 
the 6 mins to give a rate of change of +2/min and so the value 20 is stored in 
the 12:05:00 - 12:09:59 bucket.

This behaviour makes a lot of sense if you're working with large numbers and 
the Counter data type (such as for network interface traffic).  If is a bit 
more troublesome if what you are measuring is a small gauge, such as 
temperature or pages printed.

Currently there is no way to disable data normalisation.  What you can do is to 
ensure the data is logged on the time window boundary, ie at 12:00, 12:05, and 
so on.  Manual calls to RRDTool (or rateup) can do this simply by using an 
explicit time rather than 'N'.  MRTG can do it by making this particular cfg 
file be processed ever 5min boundary (use cron to force the run at */5)

Tobi has several times been requested for an RRDTool data type that is like 
'gauge but without normalisation' but I believe it is a bit harder to achieve 
programmatically than it would initially appear due to the way RRDTool works.

Steve


Steve Shipway
ITS Unix Services Design Lead
University of Auckland, New Zealand
Floor 1, 58 Symonds Street, Auckland
Phone: +64 (0)9 3737599 ext 86487
DDI: +64 (0)9 924 6487
Mobile: +64 (0)21 753 189
Email: [email protected]
 Please consider the environment before printing this e-mail 

Attachment: smime.p7s
Description: S/MIME cryptographic signature

_______________________________________________
mrtg mailing list
[email protected]
https://lists.oetiker.ch/cgi-bin/listinfo/mrtg

Reply via email to