[rrd-users] NAN and 0 in aggregate graphs

[Andy Riebs]

Hi Adam,

If you check the list archives, you'll see that the discussion of NANs 
has been thoroughly covered in the past month.

(Though, I'll confess that I don't recall anyone declaring that 
"val+NAN" is more meaningful than "NAN+NAN.")

Regards,
/andy

Adam Jacob Muller wrote:
> Hello,
> I have an application utilizing rrdtool where it is in some cases
> useful to produce "aggregate" graphs of information, this works great
> in rrdtool by just doing a CDEF=somefield_aggregate=field1,field2,+
>
> the only problem I experience here is when field1 or field2 are NAN,
> obviously when field1 and field2 are NAN the proper result is NAN, but
> if field1 is NAN and field2 is some positive value than the result
> should be field2's value.
>
> Right now I am solving this by doing something like:
>
> CDEF:field1_nz=%s,UN,0,field1,IF
> CDEF:field2_nz=%s,UN,0,field2,IF
> CDEF:field_total=field1_nz,field2_nz,+
>
> this works, but, on values where both are NAN this returns 0, it would
> be much more preferable to return NAN here
>
> Any ideas how to do this.
>
>
> -Adam
>
> _______________________________________________
> rrd-users mailing list
> rrd-users at lists.oetiker.ch
> https://lists.oetiker.ch/cgi-bin/listinfo/rrd-users
>   

-- 
Andy Riebs,     HP High Performance Computing R&D
             Technology for better business outcomes
Linux Software,      <http://www.hp.com/go/clusters>
Catalyst Program,   <http://www.hp.com/go/catalysts>
My opinions are not necessarily those of HP
(w) +1.603.884.1521, andy.riebs at hp.com
(h) andy at candooz.com              (Data are not is.)