[rrd-users] Combine average data to one CDEF

[Simon Hobson]

Hans van Kilsdonk wrote:

>This works like a charm. However, when I want to add another DEF where
>the data in the RRD file starts later the whole graph starts at that
>time. Any data before the latest start time is not shown in the graph.
>
>Is there a way to avoid this?


That's because the data is unknown - and could apply to any of the 
rrd should the collection routine stop for any reason.

   x + unknown = unknown

Thus is ANY of your values is unknown, the the total will also be unknown.

The answer is to test each one, something like this :

   0,x,x,unkn,if

Check the rrd rpn pages for more details, but x,unkn, evaluates to 
true if x is unknown, then 0,x,<expr>,if says if expr is true then 
give x, otherwise give 0.

Apply such a formula to each value, then add up the results.

One downside is that if ALL the values are unknown, then you still 
get zero, not a gap in the graph.