[ixpmanager] Spikes on MRTG graphs

[remy.guenter at swissix.ch]

We face the same problem on the historical data for month and year:

https://ixpmanager.swissix.ch/statistics/ixp

To fix this we can not just simply change some text in a log file. For us
this would mean - in my view - to dump the rrd datafiles (our installation
does not write to log files), correct the values and then restore to rrd-db.


That would require a script to find the "wrong" entries in the dumped XML,
correct them (average of entry before/after) and write back the XML to be
reloaded.

Is there a simpler procedure?

Rémy

-----Original Message-----
From: ixpmanager <ixpmanager-bounces at inex.ie> On Behalf Of Nick Hilliard
(INEX)
Sent: Saturday, 28 March 2020 1:40 pm
To: INEX IXP Manager Users Mailing List <ixpmanager at inex.ie>
Subject: Re: [ixpmanager] Spikes on MRTG graphs

Barry O'Donovan wrote on 28/03/2020 10:38:
> This is exactly as Martin suggests and also how we fix it too.

also, ixp manager puts in spike limits by counting up the total possible
bandwidth of each individual graph.  E.g. the total possible bandwidth of
the ixp would be the sum of all the edge ports, etc.  So there are checks in
there already, but even still MRTG manages to mess things up from time to
time.

Nick
_______________________________________________
INEX IXP Manager mailing list
ixpmanager at inex.ie
Unsubscribe or change options here:
https://www.inex.ie/mailman/listinfo/ixpmanager