CPU topologies mentioned twice in deployment file

Community Support
1

Using OpenNebula 6.8.0 to begin with…

I’m testing NUMA and how to configure it. If i define a topology of “THREAD” and give the VM 6 VCPUs, the resulting deployment file looks inconsistent and somewhat odd.
I have this part first:

<cpu>
     <topology sockets='1' cores='6' threads='1' />
</cpu>

And later on I have this section:

   <cpu>
          <topology sockets='1' cores='3' threads='2'/>
          <numa>
                   <cell id='0' memory='12582912' cpus='0-5'/>
          </numa>
   </cpu>

Why do I get two different sections with basically different settings?

2

@tosaraja , could you, please, share the output of onevm show -j <vmid> command, where <vmid> needs to be replaced with VM ID you are referring to in that thread?

3

It’s a mile long, so I shortened it to just contain the CPU parts:

  "CPU": "8",
  "FEATURES": {
  "IOTHREADS": "1",
  "VIRTIO_SCSI_QUEUES": "4"
 },
 "NUMA_NODE": {
   "CPUS": "40,88,4,52,20,68,28,76",
   "MEMORY": "28311552",
   "MEMORY_NODE_ID": "0",
   "NODE_ID": "0",
   "TOTAL_CPUS": "8"
 },
 "TOPOLOGY": {
   "CORES": "4",
   "PIN_POLICY": "SHARED",
   "SOCKETS": "1",
   "THREADS": "2"
 },
 "VCPU": "8",