KubeVirt

KubeVirt is a virtual machine management architecture built around Kubernetes.

Getting Started

This document contains some more background below. If you want to get started right away, make sure to continue with our Getting Started Guide.

Technical Overview

Kubernetes allows for extensions to its architecture in the form of 3rd party resources: http://kubernetes.io/docs/user-guide/thirdpartyresources/. KubeVirt represents virtual machines as 3rd party resources and manages changes to libvirt domains based on the state of those resources.

This project provides a Vagrant setup with the requisite components already installed. To boot a vanilla kubernetes environment as base for kubevirt, simply type vagrant up from the root directory of the git tree, which can be found here:

https://github.com/kubevirt/kubevirt

Once the Vagrant provisioning script has completed, run ./cluster/sync.sh to build and deploy KubeVirt specific components to the Vagrant nodes.

Note: KubeVirt is built in go. A properly configured go environment is therefore required. For best results, use this path: $GOPATH/src/kubevirt.io/kubevirt/

Associated resources

Kubernetes
Libvirt
Cockpit

Project Components

virt-api: This component provides a HTTP RESTfull entrypoint to manage the virtual machines within the cluster.
virt-controller: This component manages the state of each VM within the Kubernetes cluster.
virt-handler: This is a daemon that runs on each Kubernetes node. It is responsible for monitoring the state of VMs according to Kubernetes and ensuring the corresponding libvirt domain is booted or halted accordingly.
virt-launcher: This component is a place-holder, one per running VM. Its job is to remain running as long as the VM is defined. This simply prevents a crash-loop state.
ha-proxy: This daemon proxies connections from 192.168.200.2 to the running master node--making it possible to establish connections in a consistent manner.

Scripts