[rrd-users] Not understanding the 'create' function....

Sam Umbach sumbach at gmail.com
Fri Apr 20 16:16:04 CEST 2007

On 4/20/07, Leif Neland <leif at neland.dk> wrote:
> David Ball skrev:
> >   I feel a bit sheepish asking, but this just doesn't make sense to
> > me.  I create a test RRD as follows:
> >
> > # /usr/local/rrdtool-1.2.19/bin/rrdtool create ../tmp/testRRD3.rrd
> > --start 1167609600 --step 300 DS:bps:COUNTER:600:0:150000000
> > RRA:MAX:0.5:1:10000
> >
> > But when I do a dump of it, the first timestamp in the RRD is no where
> > NEAR what I specified in my 'create' statement.  In fact, the LAST
> > timeslot in my database is the timestamp I mentioned for the START of
> > the RRD....

An RRD contains a fixed number of data slots (representing a fixed
length of time based on the step size, steps per CDP, and number of
rows stored in each RRA).  Each time you insert data, rrdtool rolls
out old data in order to make room for the new data.  When you first
create an RRD, it is filled with Unknown data values, and these are
rolled out as you perform updates. [Strictly speaking, data isn't
rolled out every time you perform an update; data is rolled out of an
RRA every time a row boundary is passed.]

It sounds like your RRD is set up correctly and will give the results
you're expecting.  rrdtool dump is generally not used for data
retrieval, but can be helpful for diagnosing problems.  You should
retrieve data with rrdtool fetch or rrdtool graph.


More information about the rrd-users mailing list