[rrd-users] Re: wrong values when graphing with RRD

[Fernando Nieto]

Hello Alex, thank you for your answer, but I'm not completely agree with
you, let me explain. 

I have something like the following:

	DS:ds0:GAUGE:600:0:U
	DS:ds1:GAUGE:600:0:U
	RRA:AVERAGE:0.5:1:2328

Every record stores 8 days of samples collected every 300 seconds. The RRA
step is 1, so every PDP is AVERAGEed to form one CPD, that is, every value
becomes one CPD.

Now I want to get a graph like this:

	CDEF:sum=ds0,ds1,+ 
	LINE1:sum#0000FF:"max, avg, cur:" \
	GPRINT:sum:MAX:"%5.2lf %%" \
	GPRINT:sum:AVERAGE:"%5.2lf %%" \
	GPRINT:sum:LAST:"%5.2lf %%\c" \

If values are, for instance:

		...	t0	t1	t2	t3	t4	...
	ds0		0	0	10	0	0	<- CDPs
	ds1		0	0	0	0	0	<- CDPs
	sum		0	0	10	0	0	<- CDEF

If I'm not wrong, CDEF is calculated for every timestamp and its values, CF
"it doesn't matter" here, just apply the formula.

LINE1 statement should print for every timestamp, the stored values
(...0,0,10,0,0...), exactly the same as if I plot ds0 or ds1.

Now, GPRINT:sum:MAX should be 10, the MAX value of sum's series, in this
situation, I think MAX of MAXes and MAX of AVERAGEs are exactly the same. 

If you were right, maybe the MAX legend should be wrong, but not graph plot,
which shoult show the correct "AVERAGE" value. 

Anyway, the two graphs I attached are exactly the *same* plot (the same RRD
source, the same DEFs and CDEFs...), the only difference is height & width
graph option, and nevertheless both are surprisingly different!!. So, I keep
on thinking that there is something wrong with my graphs or RRD?.


cheers,
Fernando 



> -----Mensaje original-----
> De: rrd-users-bounce at list.ee.ethz.ch 
> [mailto:rrd-users-bounce at list.ee.ethz.ch] En nombre de Alex 
> van den Bogaerdt
> Enviado el: martes, 20 de abril de 2004 21:22
> Para: rrd-users at list.ee.ethz.ch
> Asunto: [rrd-users] Re: wrong values when graphing with RRD
> 
> On Tue, Apr 20, 2004 at 05:24:38PM +0200, Fernando Nieto wrote:
> >   You can see this in the attached graphs. Both shows the 
> > last 2 days data, I have only changed its width and height when 
> > I have created them.  Both graphs should show a peak of 5% near 
> > midnight, but in the "small" one is showed a 2.5% value and 
> > that's what the legend also says!! why??. If resolution was the 
> > problem, at least, the legend should show the correct value, 
> > don't you think so?
> 
> Yes, I don't think so.
> 
> You are thinking that MAX means maximum ever seen.  It is 
> not, because you are fetching the data from the AVERAGE.  You 
> are printing the MAX of the averages, not the MAX of the MAXes.
> 
> MAX(  AVERAGE(5,5) , AVERAGE(0,0) )
> 
> is different from
> 
> MAX(  AVERAGE(5,5,0,0)   )
> 
> and that is what you're reasoning fails to compensate for.
> 
> Alex
> 

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi