Check VM overrides on cluster (for example DRS or HA Restart priority).
We had similar problem. VMs which had configured overrides, didn't get VMCP config. Fortunately it can be changed with few "clicks".
