> > *> Has anyone created a tool that will create a RRD sytle xml file from the
> > *> MRTG log files? I tried but I can't seem to get it right. 
> > 
> > very good iea .... make one ... go go ... ask any questions you have I'll be
> > glad to answer ... 
> > 
> 
> Well, I can convert the data and get it graphed with MAX values but as
> soon as I do an update the graphs go hay wire and I loos the MAX,MIN, &
> AVERAGE values. That's if I don't dump core. :)
> 

As Tobi already told, some calculations need to be done.

Perhaps the following aproach will provide a "good" file so that you
can check any future development against this file:

The main problem in converting an mrtg log file to an rrd file is
that you can't copy the maxima (without some calculations).
You can however very easy import mrtg data into an rrd if you are willing
to skip the maxima.

Above is also true if you import maxima and not averages.

This means that you can create two rrds, one containing averages, one
containing maxima.

If you dump those two rrds, merge the dumps and restore the resulting
dump, you should end up with a correct database.

I wouldn't recommend this as a final solution but as written above, you
would at least get a correct database to compare your development to.


So this would come down to:

- create a database with mrtg-alike properties
  make sure to use GAUGE datasource types.
  make sure to use a start time that is about 800 days ago.
  MRTG stores 2 datasources
  MRTG stores (approx.):
      600 times  5 minute samples, time x to x+50h
      600 times 30 minute samples, time x+50h to x+14d2h
      600 times  2 hour samples,   time x+14d2h to x+64d2h
      732 times  1 day samples.    time x+64d2h to x+796d2h
  RRAs will always be from the beginning (unlike MRTG) so if you do
  not want to loose any data, you need to create your RRAs larger than
  600 rows !
- Copy this database
- Start at the bottom of the mrtg logfile and update one database with
  the average values
- Do the same to the other database and the maxima
- dump both databases
- in the "max" rrd dump:
  insert the RRAs from the "avg" dump
- alter the appropriate settings in the "max" dump, such as step time,
  datasource type
- restore the "max" dump.
(I'm not exactly sure if the averages should be merged into the maxima,
 or that it should be the other way around. Pay extra attention to this).

I may have missed something but you get the picture.
Please keep in touch, I'm very interested too.

If anybody would like to comment on this, please do.

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