[rrd-users] faking a datasource on the fly

Simon Hobson linux at thehobsons.co.uk
Wed Jul 27 20:52:01 CEST 2011

Daniele Sluijters wrote:

>This one might sound familiar to some. MRTG has the (nasty) 
>limitation that it needs to fetch pairs of data, it can't just fetch 
>1 or 3 values, they must always be grouped in logical pairs.
>This presented a problem when 
>(ucdavis.memory.memShared) had to be fetched because that value 
>isn't populated anymore. Instead however of returning 0 (which would 
>be inaccurate) or Unknown (accurate) "Snmp no such object" is 
>returned. This apparently confuses MRTG so much that even if the 
>other OID which makes up the pair . does 
>respond with sane values an RRD is created with two datasource's 
>which just always return NaN.

That's an MRTG issue, not an RRD issue. I'd suggest you should take 
that up with the MRTG people.

For my uses, I use a script to fetch the data, apply any 
manipulations required (for example, one of my sources returns an 
integer which is 10x the real value so I have to add a decimal 
place), and of course apply some sanity checks - one of which is "if 
the value isn't available, use "U for the update".

Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.

More information about the rrd-users mailing list