CNI Fails on New Node Due to x86-64-v2 CPU Incompatibility

Problem

• After adding a new compute node, services such as CNI failed to start with the following error:

Error: failed to create ambient nodeagent service: error initializing mesh dataplane: error configuring iptables: failed to execute iptables-save: exit status 127 Fatal glibc error: CPU does not support x86-64-v2

Environment

• Platform9 Private Cloud Director - v2025.4 and Higher
• Component - Kubernetes

Cause

• This issue is typically caused by using a legacy virtual CPU model (qemu64) in the Nova compute configuration (nova_override.conf), which does not expose modern CPU instruction sets required by containerized services compiled against the x86-64-v2 microarchitecture baseline.

Resolution

• Edit the nova_override.conf on the affected node to use a modern virtual CPU model, as below:

[libvirt]
cpu_models = <modern-virtual-cpu-name>

• Restart the pf9-ostackhost service to apply the changes:

$ sudo systemctl restart pf9-ostackhost

• Migrate or recreate the VMs hosted on the affected node.