This article explains why VM failover fails when an empty sizing policy (without vCPU, cores per socket, or memory defined) is used and provides guidance on how to resolve the issue.
Symptoms:
When failing over a VM between sites in VMware Cloud Director Availability:
The target VM does not retain its source configuration.
The failover process applies the target site’s default sizing policy instead of inheriting the source VM’s settings.
The failover task fails with the error:
“Virtual machine cannot support 1 CPUs using 4 cores per socket. CPU count must be a multiple of cores per socket.”
VMware Cloud Director Availability 4.7.3.1
VMware Cloud Director 10.6.1.1
This behavior occurs when a sizing policy without defined compute values (vCPU, cores per socket, memory) is attached to the VM.
During failover:
The destination site relies on the sizing policy to determine the VM’s compute configuration.
If the policy is empty, no configuration is passed, and the system falls back to the default target site policy.
This fallback can cause a mismatch in CPU/core values, leading to the failover error.
To avoid this issue, ensure that sizing policies contain explicit compute definitions.
Steps:
Create or edit the sizing policy on both source and target sites.
Define:
vCPU
Cores per socket
Memory
Assign the defined sizing policy to the VM.
Perform a Test Failover to confirm that the VM is reconfigured correctly before running a planned failover.
Conclusion
This is by design.
Sizing policies are intended to define VM resource specifications.
An empty policy does not provide any configuration details, so attaching it to a VM is not effective and will cause failover to fail.
Using explicit definitions ensures successful failover.
A Test Failover is recommended before planned failover operations.