[mrtg] Handling successive counter rollovers

[Ed LaFrance]

Hello all -

I am running mrtg 2.15.0 on linux, monitoring various ports on a Gpbs 
managed switch. I've run into an issue on the uplink port, which has 
the most traffic. The Octet counters are not 64 bit, and with a 
5-minute polling interval, it appears that multiple successive 
counter rollovers can happening in between mrtg runs. In other words 
OutOctets will roll over before a data capture, then run all the way 
up and roll over again *before* the next data capture, and so on - 
sometimes this will happen several times in succession, and the 
result is artificially low traffic reports on that interface. I've 
verified this by capturing the Octets out on that particular 
interface at one minute intervals to a log file - note the counter 
has reset by every fifth capture almost like clockwork:

1.3.6.1.2.1.2.2.1.16.24 = Counter: 275615047
1.3.6.1.2.1.2.2.1.16.24 = Counter: 1099703953
1.3.6.1.2.1.2.2.1.16.24 = Counter: 1977232053
1.3.6.1.2.1.2.2.1.16.24 = Counter: 2938038189
1.3.6.1.2.1.2.2.1.16.24 = Counter: 3871828929
1.3.6.1.2.1.2.2.1.16.24 = Counter: 425011425
1.3.6.1.2.1.2.2.1.16.24 = Counter: 1217557647
1.3.6.1.2.1.2.2.1.16.24 = Counter: 2024459227
1.3.6.1.2.1.2.2.1.16.24 = Counter: 2896359311
1.3.6.1.2.1.2.2.1.16.24 = Counter: 3803224615
1.3.6.1.2.1.2.2.1.16.24 = Counter: 384017160
1.3.6.1.2.1.2.2.1.16.24 = Counter: 1265435349
1.3.6.1.2.1.2.2.1.16.24 = Counter: 2157523201
1.3.6.1.2.1.2.2.1.16.24 = Counter: 2885065373
1.3.6.1.2.1.2.2.1.16.24 = Counter: 3742325451
1.3.6.1.2.1.2.2.1.16.24 = Counter: 210382955


I am trying a test right now in which mrtg is config'ed for the 
default nominal interval of 5 minutes, but I am actually capturing 
data every 3 minutes to avoid the successive rollovers. It seems to 
be working when I compare it to the data from my provider (who 
monitors their side of the uplink), but I suspect that this is not 
the proper way to handle things - is it? I realize that as traffic 
goes up, I'll have to increase the capture frequency further. I know 
the mrtg logfile contains a timestamp for each actual capture, so I 
guess I'm asking if my solution is 'legal'. I've been using 95.pl 
with mrtg and I want to continue doing this, so going to a different 
monitoring tool is not desired.

Thanks in advance!

Ed

===============================================================
:::::::::::::     PREMIUM COLOCATION SERVICES     :::::::::::::
===============================================================
New Media LLC                    info at servercolocation.ws
11630 Fair Oaks Blvd., #300      http://www.servercolocation.ws
Fair Oaks, CA  95628             (916) 961-0446
(866) 519-4680 Toll-Free         (916) 961-0447 Fax
=============================================================== 

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