[rrd-users] Re: Filling in NaN's?

Adam Rothschild asr at latency.net
Fri Apr 27 03:24:10 MEST 2001

On Thu, Apr 26, 2001 at 02:27:34PM -1000, Clifton Royston wrote:
>   However, I hope you realize what you are getting via your proposal is
> *not* flawed data, it's no data at all - you're only making it look as
> though you had data.  In some sense, it's "negative" data - you are
> removing the data indicating that you were unable to collect data
> values at a certain time, which could be useful data in itself. 

Agreed entirely.

> As you can tell, I'm philosophically opposed to this.

As am I.  There was a very specific and somewhat broken situation
which required this, unfortunately.  I definitely don't want to come
across as saying it's a good thing, or encouraging its widespread

>   If you only want smooth looking *graphs*, there's probably a way to
> do this with some of the CDEF expressions in the grapher, using the
> "PREV" command.  E.g. assuming orig is the def for your original data
> source, something like:
>   CDEF:smooth=orig,UN,PREV,orig,IF
> should give you a graph with flat-lines where the data is unknown, much
> like MRTG's old behavior.

Yup, that works; thanks.

What I'd really like though is a script to perform the necessary
fondlage on the RRD itself -- or more realistically, export the RRD to
XML, perform the necessary fondlage there, and then re-build the RRD.
Has anyone done this before?


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