kubevirt: Unable to launch virt-handler v0.20.4 on CoreOS with selinux

/kind bug

What happened: I started kubevirt v0.20.4 using some instructions from https://kubevirt.io/quickstart_minikube/ on a Kubernetes v1.15 cluster running on bare-metal nodes with CoreOS 1883.0.0 (~1yo …).

What you expected to happen: virt-handler fails to start with:

{"component":"virt-handler","hostname":"carmd-er-lxtb0cc47a4c27b6","level":"info","pos":"virt-handler.go:187","timestamp":"2019-09-04T06:02:51.362115Z"}
panic: failed to list labels:  

goroutine 1 [running]:
main.(*virtHandlerApp).Run(0xc0001f6000)
	cmd/virt-handler/virt-handler.go:205 +0x1ad6
main.main()
	cmd/virt-handler/virt-handler.go:457 +0x6e

How to reproduce it (as minimally and precisely as possible): I think running kubevirt on CoreOS should be enough, but hard to say.

Anything else we need to know?: It looks like the selinux/labels.go is calling semanage, and I couldn’t find it in the image.

Environment:

• KubeVirt version (use virtctl version): v0.20.4
• Kubernetes version (use kubectl version): v1.15.1
• VM or VMI specifications: N/A
• Cloud provider or hardware configuration: bare-metal x86-64 servers
• OS (e.g. from /etc/os-release): CoreOS 1883.0.0 (w/ ACS patch)
• Kernel (e.g. uname -a): 4.18.5-coreos
• Install tools:
• Others: