Issues overprovisioning CPU and memory resources because of the "disablethreshold" setting

[dnezamfar]

We have 3 x 192 core / 1.5TB of RAM hosts in a single cluster/zone. We have a lot of VM's that are created with dozens of cores configured for each, but the workloads on them are very bursty and not all cores will ever be running at the same time across all the VM's on a host. I want to overcommit VM cores and RAM by a significant factor and essentially manage the resource usage myself.

I have set the overprovisioning factor to 1000 to effectively stop it from preventing VM's from being started when there is ample system resources available. I still run into issues however because of the cluster disablethreshhold values, cluster.cpu.allocated.capacity.disablethreshold and cluster.memory.allocated.capacity.disablethreshold. I want to effectively disable this so that VM's never fail to deploy when there is significant available resources.

"Cannot allocate cluster list [1] for VM creation since their allocated percentage crosses the disable capacity threshold defined at each cluster at Global Settings Configuration [name: cluster.cpu.allocated.capacity.disablethreshold, value: 1.0] for capacity Type : CPU, skipping these clusters"

That setting is at 1.0, which is 100%, which seems to be the max? How do I set that value to 1000 or more to effectively disable it? Can I use the API? Are there any other suggestions to make sure that no setting prevents me from starting VM's because of artificial limits?

[weizhouapache]

@dnezamfar
The setting is available for global, zone and cluster.
Please check zone/cluster settings.

[dnezamfar]

In the attached image are the only settings I'm seeing related to the disablethreshold setting mentioned in the error message I'm getting from the management server. I looked in the Zone and Clusters section under Infra but I didn't see anything else I could change. I already set the overprovision setting to 1000 to prevent it from playing any role in VM creation. You can see in our cluster that we have a lot of allocated cores, more than we have physical cores, but for our environment that is OK. I'd like to effectively turn off any limits that prevent me from starting up VM's when the actual used cores is very low.