There are the assumptions made before the derived metrics calculation is initiated:
All the base metrics (in all different levels) are available (i.e, the metrics have been reported) for derived metrics calculation.
For a derived metric to be calculated, all of its base metrics should have reported a valid Number.
The limitation when using the derived metrics calculator is: A derived metric cannot be reused in another derived metric
If a derived metric has to be calculated at different levels, each metric has to be specified separately in the list of metrics under the derivedMetrics section. For example, in the following metric tree:
RAM ops : 9
Hdd ops: 3
Total ops : 2
Hdd ops: 6
Total ops: (12 + 8 = 20)
If the "Total ops" metric is required both at the queue level and server level, then the derivedMetrics section looks as follows: