[rrd-users] Question about RRDTool ..

Simon Hobson linux at thehobsons.co.uk
Thu Mar 20 22:36:01 CET 2008

Jean-Yves Avenard wrote:

>On Fri, Mar 21, 2008 at 1:00 AM, Simon Hobson <linux at thehobsons.co.uk> wrote:
>>   rrdtool graph ... -s ${Start} -e ${End} -w ${Pixels} ...
>This would show 400*5 minutes worth of data.
>I'm a bit confused with how RRDtool calculate the data quite frankly.
>In another example, I collect data every minute. The data is discrete.
>I made sure the data is entered using timestamps that are a multiple of 60.
>Using rrdtool fetch data.rrd AVERAGE, I can see the data as I've
>entered it for every minute.
>However, when I draw my graph like this:
>rrdtool graph images/speedday.png -l 0 \
>-t 'Sync Speed - DAILY' \
>--step 60 --end 1206032280 --start 1206003420 \
>-x HOUR:1:HOUR:4:HOUR:2:0:%k \

>When I run rrdtool fetch AVERAGE, I can see all my data, and I see
>that the minimum value is 13940.
>However, in the graph, it tells me that the minimum value is 13948.
>Same for the maximum, looking at the data returned by fetch, it should
>be 14048, but in the graph it shows 14036.


OK, I'll mention it again - NORMALISATION

end-start is 481 sample periods, width is 400 pixels (the default 
since you have failed to explicitly set it), therefore the data is 
normalised to fit the graph. You'll note that in the example I gave, 
I explicitly set the width of the graph to fit the number of samples 
being graphed.

If you make the width of the graph in pixels equal to the number of 
samples in the time span then the data will not be normalised and you 
will get slightly different results.

More information about the rrd-users mailing list