If you have a guest that's actively processing application traffic and you decide that it needs additional CPU cores than its original core allocation, you can increase the number of cores per slot for the guest. To do this, you'll need to temporarily disable the guest while you add cores, and then re-deploy the guest.
When you increase the Cores per Slot value on a guest, the new value must be within the total number of unused CPU cores on the slot; the system never allocates more cores per slot to a guest than the number of cores still available on the slot.
For example, suppose you have three guests on a 12-core slot, with this core allocation:
|Guest Name||Current Cores per Slot|
With a total allocation of 10 out of 12 cores allocated per slot, this means that only two cores remain unused and available to add to a guest. Therefore, if you try to add four additional cores to Guest C (from 2 to 6), the system displays a message that you are exceeding the number of cores available on the slot. In this case, the maximum number of cores per slot that you can allocate to Guest C is 4.
A potential workaround for this would be to first decrease the number of cores per slot allocated to Guest A, to free up additional cores for Guest C.
You can determine the number of cores still available on each slot by logging in to the BIG-IP Configuration utility and displaying the vCMP Guest List screen.
If you have a deployed vCMP guest, and you decide that you need to allocate more CPU cores per slot to the guest than you had originally allocated, you can increase the number of cores per slot for the guest.
To do this, you'll first need to set the guest state from Deployed to Configured. Once you've increased the number of cores per slot for the guest, you can set the guest back to the Deployed state.
Note that when switching between guest states, the process can take several minutes for the guest to shut down cleanly and restart again.