> I`m using an external program to pass the data to mrtg. The values aren`t
> counters but gauges so I`m using the Options\gauge key. But when I compare
> the data mrtg writes in the logfile (and shows in the graphs) with the
> original ones, there are differences. I had a look in the logfile and
> remarked that there are 5 columns. The first contains the time-counter and
> the next four contain values. The two last of this four columns contain the
> original values, and the two in the middle contain some similar values
> (which are shown in the graph). Here an example:
> 
> 935411465 156000 2762
> 935411465 156000 2762 156000 2762
> 935411164 12000 657 12000 657
> 935411100 45706 1062 170000 2557
> 935410800 168110 2506 170000 2557
> 935410500 152160 2343 161000 2431
> 935410200 135623 2495 183000 2721

>From your logfile I can tell that the last sample (i.e. the first line)
was taken on august 23, 12:31:06 UTC.

What happens here (at least, what I expect is happening):

You do not take samples at hh:00, hh:05, hh:10 etc but at hh:01, hh:06
etcetera. It does also take five or six seconds to start the script and
return the data to MRTG. This could also be caused by a busy computer.
MRTG stores the values internally at exactly 0,5,10,15 minutes after
the hour and interpolates the values to what it thinks is most likely
the case.

Example:
at 23:06:00 you have  6000
at 23:11:00 you have 11000
MRTG needs to store the value at 23:10:00 and interpolates it to be 10000.

The 4th and 5th column do not store averages, they store maxima. Therefore
11000 will end up there.

Perhaps you can get closer to hh:00, hh:05 etcetera but you would also
need to get rid of that six second delay in returning the value.

Regards,
Alex

--
* To unsubscribe from the mrtg mailing list, send a message with the
  subject: unsubscribe to mrtg-request@list.ee.ethz.ch
* The mailing list archive is at http://www.ee.ethz.ch/~slist/mrtg