[rrd-users] Re: Wanting an error instead of an "empty" graph

Alex van den Bogaerdt alex at ergens.op.het.net
Wed Feb 25 09:56:24 MET 2004

On Tue, Feb 24, 2004 at 07:04:04PM -0800, Sandor W. Sklar wrote:

> >> Under what conditions would a graph containing no data be of value?
> >
> > rrdtool is built to handle missing data.
> > whether it is before, after or during the time spanned by your rra's.
> > It just displays it as being NaN, which in my opinion is much better
> > than to generate errors and such.
> I'd agree with you, if there was one valid data point within the range. 
>   I'd think though, that if there were zero data points in the range 
> requested, an error would be in order.
> It doesn't matter, though, since there is nothing I can do to change it 
> without modifying the code.

If you want to know when "NaN" occurs, use a CDEF to transform these
into INF and display in bright red:


> >> if the timestamp of the last update of the rrd is
> >> earlier then one week ago, then I know that I don't have the
> >> data that I want.
> >
> > Ah, I assumed you kept on updating the rrd with nans and you wanted to
> > find
> > the last time a valid number was entered.
> Why would one update an RRD with NaNs intentionally?

- signal the viewer the connection with a device was lost
- signal the viewer the device reported it could not deliver the requested data
- signal RRDtool a reboot or other counter reset will occur

maybe more.

begin  sig
This message was produced without any <iframe tags

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