[rrd-users] Re: Strange rrdtool behaviour (IMHO)

Bartosz Oudekerk bartosz at etv.cx
Wed Jan 5 00:06:56 MET 2005

Serge Maandag is thought to have typed the 
following text on Tue, 04 Jan, 2005 at 10:09:20PM +0100:

> > The output looks like:
> > /usr/local/bin/rrdtool update /var/rrd/pingstats.rrd 
> > N:9.835:10.830:11.584:0
> > The RRD definatly gets an integer as the fourth argument, as can be
> > seen by the output from the script (which I started saving for
> > debugging purposes), but how is it then possible that I
> > get floats in the last GPRINTs?
> That's a question that should be added to the top of the rrdtool FAQ.
> Rrdtool normalizes all input.
> You are updating using N: and that is probably not on a (
> N times step seconds since 01-01 1970 00:00:00) boundary.
Ok, since 30 pings with a 2 second interval finish in about one minute,
the RRD will indeed get updated right between 2 steps (counting from
01-01-1970 00:00:00), which would explain the 6.608 for a 13% packet
loss I saw. I confirmed this by generating a 53% packet loss, GPRINT
shows 27.383, in both cases only a fraction more than half the actual
loss. If this is indeed the case it should be mostly remedied by making
the script run a minute later, or is there a better way to accomplish

> Secondly, unless the width of your graph is exactly the same 
> as the amount of RRA rows you are printing, rrdtool will average
> One or some rows over two adjacent pixels. This also averages your
> Values and that also affects your GPRINTs.

Hmmm, This is logical for the graph itsself, but IMHO shouldn't be for
raw MAX and MIN calculations as in GPRINT's. A Maximum is a maximum and
by definition shouldn't be averaged if it doesn't have to be for
viewing purposes.

BTW, am I correct in assuming that the only reason to add more RRA's
with a bigger step are to minimise CPU-time at graph generation time,
or am I missing something obvious here?

Bartosz Oudekerk
   AOLS fortune mod volume VII
   Get yourself a copy at: <URL:http://www.etv.cx/~bartosz>

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

More information about the rrd-users mailing list