[rrd-users] Re: complex CDEF questions

Alex van den Bogaerdt alex at ergens.op.het.net
Wed Jun 14 13:19:45 MEST 2006

On Wed, Jun 14, 2006 at 10:42:14AM +0000, Richard Marshall wrote:

> I'm getting large peaks in my graphs due to massive rates of change in the 
> data that is (incorrect and) very difficult to work around.

Can you solve this problem?  This may be better than to work around
the resulting damage.

> I am currently using the work around for know bad data suggested in the 
> tutorial where if the rate is above a certain value then I set the data to 
> 'unknown'. however this leaves gaps in the graph and is not really what I 
> want.

> What I would like to do is:
> If the current rate of change is substantially greater than the previous 
> value (e.g. 100+ times greater) then plot an average of the last rate and 
> the next rate of change. So that i would end up with a graph that is without 
> gaps.

I don't think you can look into the future inside a data range.  You can
however look into the past, using PREV().  Also it is possible to have a
sliding window, also using information from the past.

Modifying your request slightly, we get:

if {current rate > previous rate times 100}
then return smoothed rate
else return current rate

This means you want something like {data,PREV(data),100,*,GT} to make
a selection.  You want {data,1800,TREND} if the selector returns true,
and {data} if not.

Untested, undebugged, suggestion to try:

> P.s. it appears  the hotmail servers have difficulty in delivering to this 
> list. 4/5 i get a non deliverable.

You may want to tell the list owner (see mail headers).  Don't forget to
tell _why_ the messages are rejected or else it may be useless information.

Alex van den Bogaerdt

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://lists.ee.ethz.ch/rrd-users
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi

More information about the rrd-users mailing list