> 
> Since every data point is based on 7200 seconds (2 hours) worth of data.
> However, I keep getting the data for the 24 hour RRA. Am I
> misunderstanding something?  I did a dump and the data is definitely
> there.
>   
After some keywords from Tobi, I did some testing on this.

> 
> what fetch does it firsts selects the RRAs which cover the requested
> timespan and THEN selects the one with the requested resolution ... 
> 

It seems that this means that the start and end times should be
aligned onto exact sample-times. If not, there will be no exact match
for the desired RRA and it will select the first that contains the
entire requested time span.

So:

TIME=`date +%s`
rrdtool fetch some.rrd AVERAGE -r 3600 -e $TIME -s end-1d

is different from

TIME=`date +%s`
TIME=$((TIME/3600))
TIME=$((TIME*3600))
rrdtool fetch some.rrd AVERAGE -r 3600 -e $TIME -s end-1d

<test>
[user@home /u]$ rrdtool fetch x.rrd AVERAGE -r3600 -e924555600 -send-2h|wc -l
      5
[user@home /u]$ rrdtool fetch x.rrd AVERAGE -r3600 -e924555601 -send-2h|wc -l
     28
</test>

The second line (only a different end-time, by one second...) gives data
from another RRA. My guess is that the above assumption is correct.
If so, this is not clear from the manual (at least not for two inhabitants
of the global village) and it was not obvious to find out.

Regards,
Alex

--
* To unsubscribe from the rrd-users mailing list, send a message with the
  subject: unsubscribe to rrd-users-request@list.ee.ethz.ch