[rrd-users] RRDtool computing logic help

Simon Hobson linux at thehobsons.co.uk
Thu Apr 3 17:41:00 CEST 2014


aleluya <huangyongjian at gridsum.com> wrote:

> I am learning RRDtool, I can't find any documents that describe what would
> happen when multiple sampling points were inserted in a single step.

http://www.vandenbogaerdt.nl/rrdtool
See his tutorial on Rates, normalizing and consolidating

> Here's my test:
> 
> rrdtool create mydata6.rrd --start 1396419600 --step 4
> DS:mem:GAUGE:2000:0:4000000 RRA:AVERAGE:0.5:1:50
> 
> rrdtool updatev mydata6.rrd 1396419601:8 1396419604:14
> 
> I got the output:
> return_value = 0
> [1396419604]RRA[AVERAGE][1]DS[mem] = 1.2500000000e+001
> 
> In the example above, the step is 4 seconds, I insert two sampling points
> whose value are 8 and 14 respectively, why I got the CDP value 12.5?

You had one second @ 8 somethings/s, and 3 seconds at 14 somethings/s.

(8 * 1) + (14 * 3) = 50
50 / 4 = 12.5

Or calculating it another way :

8 * (1/4) = 2
14 * (3/4) = 10.5
2+10.5 = 12.5





More information about the rrd-users mailing list