Closed arodrime closed 5 years ago
So the issue is that we were using the average of the whole cluster and then take min/max/avg of this value, already aggregated.
Instead we need the average per node (as there are multiple CPUs) and then, from this, take the min/max/avg. This should do @rustyrazorblade.
So the issue is that we were using the average of the whole cluster and then take min/max/avg of this value, already aggregated.
Instead we need the average per node (as there are multiple CPUs) and then, from this, take the min/max/avg. This should do @rustyrazorblade.