[rrd-users] problem with values when using width

[Alex van den Bogaerdt]

On Tue, Jan 08, 2008 at 11:52:01PM +0100, Andreas Horndasch wrote:

> I checked the rrd file with 'rrdtool dump' and for the range I used, I could not find any unknown values...
> There are only values between 0 and 1.
> 
> unknown values could only appear on the start of the graph or asking for values later then the end...
> 
> If I understood you right, you calculate the result depending on the number of pixels (the width of the graph).
> Is that right?
> If it is, how could I get the best result for a graph which should be shown on a html page? (using a big width to get a good result is looking a bit strange to my eyes)

Numerical example:

If you ask to display 1600 rates, and the graph area is only 400 pixels
wide, RRDtool will have to combine 4 rates into one pixel column.
This way, 1600 rates are reduced to 400, which fit in the graph.
It's much like consolidation, only this happens at graph time.

So, if four rates are 0.1, 0.2, 0.8 and 0.9, and if the consolidation
function used is AVERAGE, then the pixel column (LINE or AREA at that
location) will show (0.1+0.2+0.8+0.9)/4 = 0.5
Indeed: you may not find 0.5 anywhere in the output of rrdtool dump.

did this help?
alex