[mrtg] Re: Spike after rebooting routers
mm at elabnet.de
Thu Apr 24 22:03:29 MEST 2003
thanks for the detailed explanation, the "overflow" that mrtg thinks happened is what I already thought the reason for these spikes is..
Maybe I should have stated that more clearly, also, for sure I can make the spike smaller with setting maxbytes, absmax etc, but all of this doesn't solve the problem at the root ;-)
I have currently about 80 hosts and 900 targets, configs are generated 90% automatically, tuning maxbytes values manually etc. is quite impossible. The maxbytes determined by reading ifSpeed is a very good thing I don't want to break up with as it affects several other things like percentages etc.
After cleaning spikes for some hours today with this horrible killspike.pl I can imagine a possible solution but my perl know-how is far away from being able to realize/change this in mrtg:
- Look on sysuptime, compare with the last timestamp in the .rrd and if it's lower than the difference, don't expect the counter wrapped and count the value received from zero.
I cannot imagine this wasn't implemented or at least discussed yet but I found nothing in the lists so far..
From: (Verstrooid) Profke [mailto:profke at easynet.be]
Sent: Thursday, April 24, 2003 5:54 PM
To: Makai József; Michael Markstaller; MRTG Mailing List
Subject: Re: [mrtg] Re: Spike after rebooting routers
maybe you can minimize the problem by setting a lower maxbytes value for
any values higher than maxbytes are discarded because they are considered
that way you should never get spikes higher then maxbytes.
----- Original Message -----
From: "Makai József" <jozsef.makai at siemens.com>
To: "Michael Markstaller" <mm at elabnet.de>; "MRTG Mailing List"
<mrtg at list.ee.ethz.ch>
Sent: Thursday, April 24, 2003 2:50 PM
Subject: [mrtg] Re: Spike after rebooting routers
> Hi Michael,
> This phenomenon is absolutely normal.
> In the router there are counters, and the mrtg checks the values in every
> After reading out the values the counters continues counting without
> their values.
> If the next readout is higher, mrtg calculates the normal difference
> the two values, and stores it.
> If the next readout is lower then the previous one, that means that the
> counter reached the 32 or 64 bit limit and continues counting from zero.
> In this case (the current value of a counter is lower then the previous
> mrtg takes into consideration the overflow of the counters.
> (mrtg can calculate only with one overflow, so if you wanted to increase
> default 5 minutes interval, you must take this into consideration!)
> After a reboot the counters are reset, and the calculated values are
> practically never real values. Either lower or higher due to the above
> mentioned. (The spikes, what you find strange.)
> I hope this helps,
> > Hi there,
> > I have a strange problem running MRTG against several cisco routers:
> > when doing a IOS-update (or only reload) I sometimes get high
> > spikes in *some* interface graphs.
> > The ifIndex doesn't change and everything works fine besides
> > from that, we're also talkking about upgrading minor
> > release-versions without any noticeable other change in SNMP
> > on the router. After the spike everything works fine again;
> > It's quite annoying because it messes up the graphs' scale
> > und my traffic totals, and fixing the spikes in the RRDs is
> > also a time consuming task..
> > Is there any known workaround out there to prevent this
> > (besides from setting AbsMax lower), either manual in advance
> > to a (scheduled) reload/upgrade or preferably something
> > generic in mrtg-config (might using snmpv2c help) ?
> > another question: is there any other tool like killspike.pl
> > which allows me to specify the time-period or somthing
> > similar) where I want to kill spikes ?
> > some of my config: mrtg-2.9.29 / RRDtool 1.0.41 on Red Hat Linux 7.3
> > thanks
> > Michael
> 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
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
FAQ http://faq.mrtg.org Homepage http://www.mrtg.org
More information about the mrtg