[rrd-users] "not a simple integer" error with DERIVE ds

Alex van den Bogaerdt alex at ergens.op.het.net
Sat Jun 9 01:26:25 CEST 2007

On Thu, Jun 07, 2007 at 09:13:58PM +0530, Nirnimesh wrote:
> I get "not a simple integer" error with the DERIVE ds. I looked up the
> archives and as far as I can understand, the only solution I can find is to
> switch to integers. However, I do have valid reasons why I cannot, without
> incurring too much pain. I expected this to be handled in a neater way. Am I
> missing something?

It's all about resolution of the data.  There's a limited amount
of significant digits available, and choices had to be made. 

A workaround: You can multiply the input by a certain amount (e.g. 1,000
, or 1,000,000) and undo this using a CDEF.  This is how I monitor some
values; it seems to work well.  It also works out when you do the math
yourself, no problem:

Time 't'     123.500
Time 't+300' 124.100
-------------------- diff
300 seconds    0.600
0.6/300 = 0.002

Time 't'     123500
Time 't+300' 124100
-------------------- diff
300 seconds     600
600/300 = 2

Now divide by 1000 using a CDEF, and you get 0.002 again.

Alex van den Bogaerdt

More information about the rrd-users mailing list