[rrd-users] MAX values dependant on width of graph

[Simon Hobson]

Fabien Wernli wrote:

>just to make some noise, I am still looking for an application of using the
>minimum of a MAX rra ;-)

How about, you want to know if a value failed to reach a threshold in 
any sample period ? So if you take the minimum of a max RRA then you 
can see immediately if that was the case.

Now, trying to think of an application for that ... err still 
thinking ... err how about this. You have a machine that runs some 
sort of cyclic operation. For the operation to be successful, the 
pressure (or temperature) in <something> must reach a set minimum 
value at least once during the cycle but as long as it reaches that 
then the actual pressure (temp) doesn't matter. It costs money to 
apply more than the minimum, so you don't want to just dial in a 
large safety margin. We can assume that a machine controller is 
running the show, but you monitor the pressure (or temperature) and 
by plotting "min of max" you can look at the trend of the "trigger 
condition".

OK, so it's rather contrived !