[rrd-users] Question about averaging daily ; but with 5 min. steps

Jean-Yves Avenard jyavenard at gmail.com
Tue Dec 15 13:00:45 CET 2009


I got a bit closer to what I wanted to do today ...

2009/12/15 Jean-Yves Avenard <jyavenard at gmail.com>:
> However, if I use 5 minutes average (the same calculation as performed
> by the electric meter) the amount of energy put back in the grid is:
> (1000 - 500) * 10 / 60 * 24 / 1000 = 2kWh per day.
> \
>   CDEF:cexport=csolar,UN,0,csolar,IF,ctotal,-,0,MAX \
>   CDEF:cexport30=csolar30,UN,0,csolar30,IF,ctotal30,-,0,MAX \
>   CDEF:cexporttot=solar30,UN,0,solar30,IF,total30w,UN,0,total30w,IF,-,0,MAX,$duration,*
> So back to my original question:
> How do I graph with RRD ; a daily average made on a 5 minutes average
> of a substraction of 2 values of two different RRD input.
> And is this even possible currently ?

I calculated as follow:
    CDEF:cexport30b=COUNT,$res,300,/,%,0,EQ,cexport30,$res,TRENDNAN,UNKN,IF \

Basically, I look at the position of the value in the dataset ; if it
doesn't fall on a multiple of 5 minutes, I push UNKNOWN, if not ,
using TRENDNAN, I calculate the average at this point of the last 5
minute and push it on the stack.

Now, I can calculate the maximum and the minimum properly...
Unfortunately, I still can't plot properly because rather than putting
UNKNOWN, I should be putting nth-time the average just calculated.
So if calculating an hourly average, I should be duplicating 12 times
the 5 minutes average.

Unfortunately, the RPN stuff doesn't give me that much leeway...

Does anyone have any other ideas on how to achieve what I've been talking about?


More information about the rrd-users mailing list