Warning
If this MOP is not followed properly to increase the vCPUs, the performance improvements may not be attained and there can be an impact on existing call capacity.
SBC Core release 8.1 includes a feature that extends the use of hyper-threading to SBC SWe when it is installed on either the VMware or KVM Hypervisor platform. To take advantage of the performance improvements provided by hyper-threading, you must increase (double) the number of vCPUs configured in the VM prior to software upgrade. When upgrading, modify the upgrade process to incorporate the following additional steps:
- On the VMware platform only, before beginning the upgrade, disable the CPU reservation check inside the guest by renaming the file:
vmware-toolbox-cmd
For example, issue the following command:
mv /usr/bin/vmware-toolbox-cmd /usr/bin/vmware-toolbox-cmd.bak
- For either platform, before starting the upgrade procedure, shutdown the standby VM instance and double the number of vCPUs specified for the VM. Refer to the procedure you used creating a VM that is appropriate for your specific deployment:
(VMware) Creating a New SBC SWe VM Instance with VMXNET3 (step 7)
(VMware) Creating a New SBC SWe VM Instance with Direct IO Passthru (step 7)
(KVM) Creating a New SBC SWe Instance on KVM Hypervisor (step 6)
(KVM) Creating a New SBC SWe Instance with PCI Pass-Through Device (step 6) - After increasing the number of vCPUs, power on the standby instance with the increased number of vCPUs in the same release.
- Switch over the active instance, so that the standby becomes the new active instance. Don't wait for the standby instance to come up. It may not come up due to the resource mismtach.
- Shutdown the new standby instance and double its number of vCPUs, as done earlier in step 2.
- Bring up (power on) the new standby and wait for the active and standby nodes to sync.
- Upgrade the standby node to SBC Core release 8.0.
- Once the standby node is upgraded, begin the upgrade of the active node to SBC Core release 8.0. The standby node becomes the new active node as a result of auto-switchover.
- After the system upgrade completes, wait for the sync between nodes to be completed.