[mrtg] MRTG-bugs

Steve Shipway s.shipway at auckland.ac.nz
Thu Dec 18 00:22:30 CET 2008

My guess is that there are 3 issues here:


1. Data time window normalisation.

MRTG normalises the data to exact 5-min windows.  If your data come in at
any time other than exactly on the 5-min boundary, the value is adjusted to
be what it would have been at this time, assuming a straight line joining
the last known data point and this one.  This is sensible when graphing
network data but can result in unexpected changes when graphing other
low-valued gauge data.  The log file does not show the data/time that MRTG
obtained but the normalised data.


2. 1024/1000 confusion.

1KB=1024B but 1kB=1000B.  So how do you represent it?  Make sure that both
things you are comparing are using the same size of K. you might like to get
the raw numerical data for comparison instead - if your router reports in K
but your graph in k then the graph will appear to show larger values.


3. Comparing different time windows

If you are reading a 5-min average use from a calculation made by your Cisco
and comparing it to MRTG then it will be different as they are working on
different time windows.  Of course they will be approximately the same but
not identical.


Also note that you sent us the HTM file for the page, not the graph images
(however I've seen one of these already).  Remember a MRTG graph will show 2
variables, one given as a green area and one as a blue line.




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/mrtg/attachments/20081218/22453275/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3133 bytes
Desc: not available
Url : http://lists.oetiker.ch/pipermail/mrtg/attachments/20081218/22453275/attachment.bin 

More information about the mrtg mailing list