review the metrics calc for A/P and A/A deployments for cpu, memory usage
kami619 opened this issue · comments
When we made a change to how we are capturing the metrics when two clusters have active load, we might have skewed the calc for A/P setup. We should re-look at this approach and determine if the current method would work for both deployment types or we should rather not halve the CPU seconds spent even in the A/A tests as in A/A we need to set number of cpus to a number that is able to handle load even if the second site fails therefore we should request the same amount of cpus in both A/A and A/P.
A similar opportunity for improvement in the metrics actions was identified in a prior logged issue - #925