[rrd-users] Getting NaN in CDEF with LIMIT but the DEF has a validnumber

[Alex van den Bogaerdt]

I  was typing my reply when Simon's came in. I believe it may be something different than normalization.

----- Original Message ----- 
From: "Tom Jefferson" <tjefferson2001 at hotmail.com>
To: <alex at vandenbogaerdt.nl>; <rrd-users at lists.oetiker.ch>
Sent: Wednesday, December 28, 2011 6:00 AM
Subject: RE: [rrd-users] Getting NaN in CDEF with LIMIT but the DEF has a validnumber



> Thanks for the suggestion, that is indeed the issue. So when using a LIMIT in a CDEF should
> you allow for a little overage on your range?  Is this a common issue?   

I'm not a mathematician, but my gut feeling says that comparing floating point values (google for it) is not just about the equal sign. Google for "comparing floating point values" and you'll find plenty of information.

I'm not sure if this is something the user should compensate for, or if RRDtool needs to be altered.