Hi all,
I'm working on a virtual networking lab leveraging ESXi nested virtualization technology. Host is a Cisco UCS server based on dual socket Intel Xeon E5-2690 equipped with 128 GB RAM and running ESXi 6.0. Basically ESXi 6.0 supports a VM (16 vCPU, 64GB RAM assigned) running unetlab software. Internally unetlab, based on linux 64-bit kernel, supports multiple VMs (each of them implementing a network node - basically a "router") leveraging QEMU/KVM technology.
To enable Virtualized Hardware-Assisted Virtualization (VHV) for the VM running on ESXi I set vhv.enable = "true" in the .vmx configuration file.
nested VM instances running router OS have problems when multiple-vcpu are assigned to them (basically OS crash and try to reboot). No issues seen however when configured to run with just 1 vcpu assigned.
Note that when unetlab runs on bare metal server no issue is seen even with multiple vcpus assigned to router VMs running on top QEMU/KVM.
Any idea to troubleshoot the issue above ?
thanks for your help !