[rrd-users] RRDtool spread values by time on rrd, how to turn off this behavior?

Simon Hobson linux at thehobsons.co.uk
Mon Apr 16 20:22:22 CEST 2012

Daniel Hilst wrote:

what must be the most frequently asked question on this list

>2) The feed it with:
>rrdtool update test.rrd 1334578211:10
>rrdtool update test.rrd 1334578221:0
>I get
>1334578210 9.000000
>1334578220 1.000000

Terse answer: This is normal behaviour, it is **EXACTLY** how RRD is 
intended to work, and no you can't turn it off.

Garrulous answer:
RRD is designed to do one thing well, normalise and store data at 
specified resolutions, in an economic way. It is not a general 
purpose database - if you want to store arbitrary data for arbitrary 
times and get the original data out then you should use a regular 
**ALL** data is normalised and consolidated according to the 
parameters used when creating the database. With some care* it is 
possible to make the normalisation a null operation, and using a 
consolidation process of one data point consolidated into one 
consolidated data point will make the consolidation a null operation 
as well - but both processes are still performed.

Alex has an excellent set of tutorials at
In particular, see "Rates, normalizing and consolidating"

* To make normalisation a null operation, you must supply data values 
every step time, and they must be timestamped **exactly** on the step 
time boundary. These boundaries are always an integer multiple of 
step time since unix epoch (midnight, Jan 1, 1970 UTC) and this 
cannot be changed.
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

More information about the rrd-users mailing list