[mrtg] Re: 8Gbps stream creates incorrect graphs- rateup problem?

[Chris Robb]

Someone suggested that it might be related to a counter bug in version of
MRTG between 2.9.17 and 2.9.19, so I upgraded to 2.9.22. The problem is
still occuring. On one of the graphs, I removed the "offending lines",
deleted the pngs, and reran MRTG. The graphs looked good. I then added the
lines back into the config, deleted the pngs and it went back to the
constant 4Kbps stream.

I made some headway on a good diagnosis. I removed all but one of the
offensive lines. I picked one large number and moved it around to each of
the input places after the epoch (in, in max, out, out max). The only one
that exhibited problems was the maximum inbound value- the third value in
the line:

1033779900 1018626663 805306366 1018626663 1020308395
                          ^ problem value

So I started changing this value until I found which number was
problematic. All values greater than 805306367 screwed up the graphs.

To eliminate the possibility of it being a transition problem between the
previous or subsequently measured values (or any of the other numbers in
the line), I moved the value 805306368 around into several different lines
within the config. It screwed them up, soo. But 805306367 worked in all
those places. So, it's not a problem with transition between one measured
value and the other.

It's also not a problem in any of the other values in the line. I'd say we
have a variable somewhere that needs adjustment.

This value is in bytes per second. It equals approximately 6.442Gbps. It
doesn't appear that this would be a data type issue given that 805306367
Bps equates to essentially the same value- 8 bits less.

I've never looked at the MRTG code before today, but I'm going to start.
If someone more familiar with the path of this value through rateup can
also have a look, I'd love it.

-Chris


Chris Robb
Indiana University Global NOC
Abilene/TransPAC Network Engineer
chrobb at iu.edu  Desk: 812-855-8604 Cell: 812-325-8199

On Sat, 5 Oct 2002, Alex Nguyen wrote:

> Hi,
>
> I have this problem with a 2Gbps stream (an aggregation of 64bits counters
> of 2 GigE links).
>
> With the version 2.9.7, I have the same graph without any error message.
>
> With the lastest version of MRTG, 2.9.22, I have the same graph and
> the following error message in MRTG :
>
> line 356:              warn "ERROR: Target[$rou][$mode] '$$rcfg{target}{$rou}' evaluated to '$data' instead of a number\n";
>
> I think, the problem is not the interval value (32bits interger), but the
> last readed value is greater than what MRTG can handle.
>
> It anyone know what is the limit of the last readed value that MRTG can
> handle ? (the value that returned by the device when MRTG is polling)
>
> Regards.
>
> A.



--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://www.ee.ethz.ch/~slist/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://www.ee.ethz.ch/~slist/lsg2.cgi