[ixpmanager] IXPM 5.3 Regenerate Yearly Traffic Graph

[remy.guenter at swissix.ch]

Ciao Nik

I have correced those values from the sequence in ixp-0001-bits.rrd:

                <cf>MAX</cf>
                <pdp_per_row>288</pdp_per_row> <!-- 86400 seconds -->

and they are now normal values. Still the graph is showing them. Therefore I wonder when this graph is created. Can you point me to the code that creates this yearly graph?

Thanks, Rémy

-----Original Message-----
From: Nick Hilliard (INEX) <nick at inex.ie> 
Sent: Friday, 3 April 2020 12:52 pm
To: remy.guenter at swissix.ch
Cc: INEX IXP Manager Mailing List <ixpmanager at inex.ie>
Subject: Re: [ixpmanager] IXPM 5.3 Regenerate Yearly Traffic Graph

Hi Remy,

RRD stores separate archives for maximum and average time series, calling them "MAX" and "AVERAGE" respectively;  the API pulls data from the AVERAGE archive.

The peaks that are showing up in your graph are dark green, which means that they use data from the "MAX" archive.

If you manually run rrdtool dump on that RRD, you should run a text search for the line which starts "<cf>MAX</cf>".  The spike values will be contained in the following <database> archive.  You will probably need to manually edit these, then recreate the RRD archive.

Removing spikes from RRD files is a bit of a nuisance.

Nick

remy.guenter at swissix.ch wrote on 03/04/2020 11:29:
> Dear Nick
> 
> But the spikes are not visible in the Log/json API version (see the 
> two calls in the mail).
> Does this mean PNG and LOG (JSON) uses different sets in the API.
> 
> Thanks, Rémy
> 
> -----Original Message-----
> From: Nick Hilliard (INEX) <nick at inex.ie>
> Sent: Friday, 3 April 2020 12:00 pm
> To: INEX IXP Manager Users Mailing List <ixpmanager at inex.ie>; 
> remy.guenter at swissix.ch
> Subject: Re: [ixpmanager] IXPM 5.3 Regenerate Yearly Traffic Graph
> 
> remy.guenter at swissix.ch wrote on 03/04/2020 10:20:
>> What am I missing here? Do in need to force recreation of the yearly
> graph?
> 
> RRD stores multiple datasets.  Probably what's happened here is that 
> you've removed the spikes from one of the datasets but not the others.
> 
> Nick
> 
>