I am evaluating vOneCloud 1.6/1.8 and plan to buy production support but I need to show this working fully.I I love it the software but one problem that I cand find a answer to is the CPU overcommit issue. As you can see below my Allocated CPU’s are maxed out while my real CPU (shown by vOne) are largely idle. in vCenter as well these hosts are idle. They each have 4 pCPU’s and 6 cores each so there is plenty of cycles but for some reason OpenNebula is only allowing me to spin up 72 or so 1 vCPU machines before I hit the max count. The VM’s are getting spread across all the hosts correctly via DRS but im at a loss as to what I need to fix to get this working correctly in OpenNebula.
Have you tried the RESERVED_CPU and RESERVED_MEM attributes in the host? If so, let me know if it is not working. Please note that you should set negative values for these attributes to provide overcommitment rather than capping the host capacity.
For instance, in the attributes host you can add what’s described in the picture below to get an overcommitment of 1 CPU and 1GB of RAM:
Overcommitment for vertical resizing is not currently supported. I’ve added a note to keep this in mind for 5.0 if the overcommitment enhancement finally makes it into the new release: