OpenNebula generates template with 12 sockets instead of 12 cores
|Affected Versions:||OpenNebula 4.10|
When instantiating instances with more then 1 vcpu, qemu-kvm runs with...
X being the vcpu assigned.
It seems to be fine in Linux guest as it can still uses more than 1 vcpu, however in Windows guest it only use 2 cores/sockets/processors. I created a VM with 16 vcpu with 16 real cpu on a dual socket server, it runs with...
And Windows guest can only see 2 processors despite that the license is permitted to have more than 2 sockets.
Also, I'm concerning if that is NUMA-aware.