hyperconverged-cluster-operator

README

Hyperconverged Cluster Operator

A unified operator deploying and controlling KubeVirt and several adjacent operators:

• Containerized Data Importer
• Scheduling, Scale and Performance
• Cluster Network Addons
• Node Maintenance

This operator is typically installed from the Operator Lifecycle Manager (OLM), and creates operator CustomResources (CRs) for its underlying operators as can be seen in the diagram below. Use it to obtain an opinionated deployment of KubeVirt and its helper operators.

In the HCO components doc you can get an up-to-date overview of the involved components.

Installing HCO using kustomize (Openshift OLM Only)

To install the default community HyperConverged Cluster Operator, along with its underlying components, run:

$ curl -L https://api.github.com/repos/kubevirt/hyperconverged-cluster-operator/tarball/main | \
tar --strip-components=1 -xvzf - kubevirt-hyperconverged-cluster-operator-*/deploy/kustomize

$ ./deploy/kustomize/deploy_kustomize.sh

The deployment is completed when HCO custom resource reports its condition as Available.

For more explanation and advanced options for HCO deployment using kustomize, refer to kustomize deployment documentation.

Installing Unreleased Bundle Using A Custom Catalog Source

Hyperconverged Cluster Operator is publishing the latest bundle to quay.io/kubevirt before publishing tagged, stable releases to OperatorHub.io.
The latest bundle is quay.io/kubevirt/hyperconverged-cluster-bundle:1.10.0-unstable. It is built and pushed on every merge to main branch, and contains the most up-to-date manifests, which are pointing to the most recent application images: hyperconverged-cluster-operator and hyperconverged-cluster-webhook, which are built together with the bundle from the current code at the main branch.
The unreleased bundle can be consumed on a cluster by creating a CatalogSource pointing to the index image that contains that bundle: quay.io/kubevirt/hyperconverged-cluster-index:1.10.0-unstable.

Make the bundle available in the cluster's packagemanifest by adding the following CatalogSource:

cat <<EOF | oc apply -f -
apiVersion: operators.coreos.com/v1alpha1
kind: CatalogSource
metadata:
  name: hco-unstable-catalog-source
  namespace: openshift-marketplace
spec:
  sourceType: grpc
  image: quay.io/kubevirt/hyperconverged-cluster-index:1.10.0-unstable
  displayName: Kubevirt Hyperconverged Cluster Operator
  publisher: Kubevirt Project
EOF

Then, create a namespace, subscription and an OperatorGroup to deploy HCO via OLM:

cat <<EOF | oc apply -f -
apiVersion: v1
kind: Namespace
metadata:
    name: kubevirt-hyperconverged
---
apiVersion: operators.coreos.com/v1
kind: OperatorGroup
metadata:
    name: kubevirt-hyperconverged-group
    namespace: kubevirt-hyperconverged
---
apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
    name: hco-operatorhub
    namespace: kubevirt-hyperconverged
spec:
    source: hco-unstable-catalog-source
    sourceNamespace: openshift-marketplace
    name: community-kubevirt-hyperconverged
    channel: "1.10.0"
EOF

Then, create the HyperConverged custom resource to complete the installation.
Further information about the HyperConverged CR and its possible configuration options can be found in the Cluster Configuration doc.

Using the HCO without OLM or Marketplace

Run the following script to apply the HCO operator:

$ curl https://raw.githubusercontent.com/kubevirt/hyperconverged-cluster-operator/main/deploy/deploy.sh | bash

Developer Workflow

If you want to make changes to the HCO, here's how you can test your changes through OLM.

Build the HCO container using the Makefile recipes make container-build and make container-push with vars IMAGE_REGISTRY, REGISTRY_NAMESPACE, and CONTAINER_TAG to direct it's location.

To use the HCO's container, we'll use a registry image to serve metadata to OLM. Build and push the HCO's registry image.

# e.g. quay.io, docker.io
export IMAGE_REGISTRY=<image_registry>
export REGISTRY_NAMESPACE=<container_org>
export CONTAINER_TAG=example

# builds the registry image and pushes it to 
# $IMAGE_REGISTRY/$REGISTRY_NAMESPACE/hco-container-registry:$CONTAINER_TAG
make bundleRegistry

Create the namespace for the HCO.

kubectl create ns kubevirt-hyperconverged

Create an OperatorGroup that watches all namespaces.

cat <<EOF | kubectl create -f -
apiVersion: operators.coreos.com/v1
kind: OperatorGroup
metadata:
  name: hco-operatorgroup
  namespace: kubevirt-hyperconverged
spec: {}
EOF

Create a CatalogSource and a Subscription.

If OLM Operator and Catalog Operator run in a namespace different than openshift-marketplace, replace openshift-marketplace with it in the CatalogSource and Subscription below.

cat <<EOF | kubectl create -f -
apiVersion: operators.coreos.com/v1alpha1
kind: CatalogSource
metadata:
  name: hco-catalogsource
  namespace: openshift-marketplace
spec:
  sourceType: grpc
  image: $IMAGE_REGISTRY/$REGISTRY_NAMESPACE/hco-container-registry:$CONTAINER_TAG
  displayName: KubeVirt HyperConverged
  publisher: Red Hat
EOF

cat <<EOF | kubectl create -f -
apiVersion: operators.coreos.com/v1alpha1
kind: Subscription
metadata:
  name: hco-subscription
  namespace: kubevirt-hyperconverged
spec:
  channel: "1.10.0"
  name: community-kubevirt-hyperconverged
  source: hco-catalogsource
  sourceNamespace: openshift-marketplace
EOF

Create an HCO CustomResource, which creates the KubeVirt CR, launching KubeVirt, CDI, Network-addons, VM import, TTO and SSP.

kubectl create -f deploy/hco.cr.yaml -n kubevirt-hyperconverged

Create a Cluster & Launch the HCO

1. Choose the provider

#For k8s cluster:
$ export KUBEVIRT_PROVIDER="k8s-1.17"

#For okd cluster:
$ export KUBEVIRT_PROVIDER="okd-4.1"

1. Navigate to the project's directory

$ cd <path>/hyperconverged-cluster-operator

1. Remove an old cluster

$ make cluster-down

1. Create a new cluster

$ make cluster-up

1. Clean previous HCO deployment and re-deploy HCO
   (When making a change, execute only this command - no need to repeat steps 1-3)

$ make cluster-sync

Command-Line Tool

Use ./cluster/kubectl.sh as the command-line tool.

For example:

$ ./cluster/kubectl.sh get pods --all-namespaces

Deploying HCO on top of external provider

In order to use HCO on top of external provider, i.e CRC, use:

export KUBEVIRT_PROVIDER=external
export IMAGE_REGISTRY=<container image repository, such as quay.io, default: quay.io>
export REGISTRY_NAMESPACE=<your org under IMAGE_REGISTRY, i.e your_name if you use quay.io/your_name, default: kubevirt>
make cluster-sync

oc binary should exists, and the cluster should be reachable via oc commands.