This is how i did it. There is a lot of work to get it back but i now have all my production vrops running at 7.0 using the below to do the allocation model.
For me it is 10 time better that 6.6.1 as the allocation model was for me broken as it used average vcpus assigned to VMs and and actual and only updated every 24 hours so if you added 100 more vcpus to VMs it took 30 days to fully reflect the 100 extra . Using the above in the link we how use actual vcpu allocated and it is refreshed every 10 mins . It is great when we have Ansable do rest API calls to vrops for placement based on the super metrics.
With supermetrics it is possible to replicate the vm_remaining concept.
With following fomula you can compute the vm_remaining for each resource type (cpu, mem) and then take the minimum: vm_remaining_per_resource = (usable_capacity * allocation_ration - amount_of_resources_occupied_by_deployed_vms) / vm_profile_size
This is how i did it. There is a lot of work to get it back but i now have all my production vrops running at 7.0 using the below to do the allocation model.
For me it is 10 time better that 6.6.1 as the allocation model was for me broken as it used average vcpus assigned to VMs and and actual and only updated every 24 hours so if you added 100 more vcpus to VMs it took 30 days to fully reflect the 100 extra . Using the above in the link we how use actual vcpu allocated and it is refreshed every 10 mins . It is great when we have Ansable do rest API calls to vrops for placement based on the super metrics.
