[rrd-users] Managing unknown values

Fridtjof Busse fbusse at gmx.de
Mon Aug 9 09:59:41 CEST 2010


Hi,

I hope this is not too much of an FAQ:
I'm using rrdtool to graph the temperature readout of a sensor.
The step of the DB is 60 seconds (to get a high resolution), the cronjob
runs every minute as well.
RRA looks like this: RRA:temp:GAUGE:600:U:U RRA:AVERAGE:0.5:1:2880

Unfortunatly, that sensors is sometimes slow to respond and returns no
value, in which case my scripts returns "U".
This is an example of what is written with rrdupdate:
25.4
U
25.4
25.4

After this, the RRD itself contains this
2.5381565507e+01
1.0256079733e+00
2.0670413320e+01
2.5400000000e+01

That creates an "ugly" spike in the graph.

What is the recommend way to make sure that an unkown value does not
have such a huge impact?
Basically I'd like values written to the RRD as unknown to be ignored or
interpolated from valid data before and after the unknown readout.

Thanks.
-- 
Fridtjof Busse



More information about the rrd-users mailing list