[rrd-users] Re: RRDtool lying about data values

PAUL WILLIAMSON pwilliamson at mandtbank.com
Fri Jan 2 16:01:37 MET 2004


I would try two things, as I'm not exactly sure of what would happen.

Try to start up MRTG with an interval of 1 minute for a new target.
Let MRTG create the RRD and see what RRAs it creates.  I *think* 
it will still create a 5 minute RRA, not an interval lower than that.

Create a RRD with an RRA of 1 minute and have Options: perminute 
in the MRTG config.  This will force MRTG to not do any 5 minute 
normalization.

Do what you thought (dump the RRD, create a new RRD with a 1 
minute RRA), and let MRTG go on it's way.  

I think #2 will work as expected, but I don't have the time to test the 
theory.

Paul

>>> <jon.hartman at verizon.com> 01/02/04 09:55AM >>>
That fixed the problem; thank you, so much. I don't have as many Y-axis
markers, which kind of stinks, but at least the graph is accurate.

Something that Paul Williamson said echoed a question I'd been pondering
as well. Since I've converted to rrd's from the MRTG flat-file which had
5-minute resolution, am I still only getting 5-minute average visibility,
even though I've set my polling interval to every minute and modified the
crontab appropriately?

I tried the following test:
-View the graph, note the current values.
-Delete all the png's in the /tmp directory.
-Wait 60 seconds and repeat.

Sure enough, he's right. The current values only changed every 5 minutes.
So, what do I have to do? Write a script that will export all the values
into XML, create new rrd's with a 1-minute resolution and then duplicate
the values another 4 times to maintain the history?

Thanks in advance,

-Jon Hartman


-----Original Message-----
From: Pavel Ruzicka [mailto:pavouk at pavouk.org] 
Sent: Friday, January 02, 2004 3:06 AM
To: Jonathan M. Hartman; pwilliamson at mandtbank.com;
rrd-users at list.ee.ethz.ch 
Subject: Re: [rrd-users] Re: RRDtool lying about data values


Hi,

> I appreciate the quick response, but I don't believe that to be inline 
> with what I've read. As I understand it, the MRTG 5 minute limitation 
> only applies if one is using MRTG to store the data. I've heard that 
> once you apply the logformat command, you can set it as low as one 
> minute.
>
> Can anyone corroborate this and if so, what would be the cause of the 
> inconsistency?

I was this problem too. This is bug in --alt-y-mrtg option in RRDTool.
This bug is in --alt-y-grid option too, but there are not much visible.
This bug is in all versions of rrdtool (tested to 1.0.45) where this
command 
is implemented. With some values is problem visible more, with other
values is visible less, but it's still here.

Solution of this problem is to edit 14all.cgi script and remove from them
"--alt-y-mrtg" option in rrdtool calling. Best solution is to repair
computing routines in RRDTool code and sending patch to author, but I'am
not programmer :-(

I report this before few months, but nobody have time to repair it :-(

Pavouk

> Thanks again for the help,
>
> -Jon Hartman
>
> -----Original Message-----
> From: PAUL WILLIAMSON [mailto:pwilliamson at mandtbank.com] 
> Sent: Wednesday, December 31, 2003 8:33 PM
> To: rrd-users at list.ee.ethz.ch; Jonathan M. Hartman
> Subject: Re: [rrd-users] RRDtool lying about data values
>
>
> Jon,
>
> If you're still using MRTG to poll the values,
> the minimum you can do is still 5 minutes.  The
> reason is that MRTG still created the RRAs, which
> are still 5 minute samples.
>
> Use something like cacti or a roll your own poller
> and build an MRTG-like utility to stuffthe data into
> the RRD database.
>
> Paul
>
> >>> <jon.hartman at verizon.com> 12/31/03 16:24 PM >>>
>
> I'm using using the following software to monitor link utilization on 
> network devices:
>
> MRTG 2.9.17-3
> RRDtool 1.0.45
> 14all.cgi
>
> I had started using MRTG alone, then converted to rrdtool and 
> 14all.cgi. I also dropped my polling interval from 5 minutes to 1. 
> When I look at the graphs, the values listed under them don't mesh 
> with the data points on the curve. As one can see in the example: 
> http://home1.gte.net/jodih/graph.png. The maximum out is stated to be 
> 113.028Mbps, while the graph clearly shows it to be more about 
> 150Mbps.
>
> Is there some sort of math error here?
>
> Thanks in advance,
>
> -Jon Hartman



--
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://www.ee.ethz.ch/~slist/rrd-users 
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi 

--
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://www.ee.ethz.ch/~slist/rrd-users
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi



More information about the rrd-users mailing list