[rrd-users] Re: MAX doesn't work as expected: peaks are filtered
rrd-users01 at ghezzi.net
Tue Aug 17 14:09:51 MEST 2004
My point is that "step" means "precision" only when using "MAX": by
using smaller step values, the average value computed for the step (the
PDP) is closer to the maximum value within the step.
If you use "AVERAGE", on the contrary, you always get the exact average,
regardless of the step size: if a step of 10 gives you the values 150
and 110, then a step of 20 gives you 130, which is the average over the
So, you see, the "step" parameter has different semantics when used with
different consolidation functions.
I find this misleading. Doesn't anybody agree? Don't you think this
could be easily changed, so to produce exact values for MAX/MIN also?
Alex van den Bogaerdt wrote:
> On Mon, Aug 16, 2004 at 10:40:47AM +0200, Cristian Ghezzi wrote:
>> Thank you for your prompt answer and sorry for sounding so negative.
>> I actually like rrdtool and I'm using it in my work.
>> Still I find it counter-intuitive sometimes, as shown in my example.
>> I would expect rrdtool to gather different kinds of PDPs to be used
>> by different kinds of CFs, so that I could have one database with 2
>> RRAs, one of them using AVERAGE and the other one using MAX, with a
>> step of 300, and everything working properly because one PDP would
>> hold the average value to be fed to the first RRA, while the other
>> PDP would hold the maximum value to be fed to the second RRA.
>> Using "--step 1" would not be the best solution, but just a hack.
> You want one-second precision, you are going to have to use "--step 1".
> That is not a hack, that is a configuration item.
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help mailto:rrd-users-request at list.ee.ethz.ch?subject=help
More information about the rrd-users