Skip to content

Latest commit

 

History

History
170 lines (125 loc) · 6.26 KB

USERGUIDE.md

File metadata and controls

170 lines (125 loc) · 6.26 KB
Raw

User Guide

이미지설명

Kubevirt-Image-Service (KIS) has 3 custom resources to manage images and volumes.

  • VirtualMachineImage : imports a qcow2 image from external sources like HTTP, S3, and local path to K8s cluster. Imported image is saved as read-only PVC in K8s cluster and will be used to create volume for VMs. Currently HTTP is only supported import option.
  • VirtualMachineVolume : creates a volume which will be used by VM from an image. Different from read-only image, created volume is able to write data. Only changed data between each volume is stored by utilizing snapshot and restore feature of CSI (Container Storage Interface). By this way, user can manage storage capacity efficiently.
  • VirtualMachineExport : converts a volume to a qcow2 file and exports it to external destinations. Currently export to local destination is only supported import option.

Before you begin

  • deploy K8s cluster through minikube, kubeadm, kubespray, or other methods
    • If K8s version is below 1.17 make sure volume snapshotting feature is enabled. Set the following flag on the API server binary: --feature-gates=VolumeSnapshotDataSource=true. Check this article for more details.
  • deploy kubevirt

A CSI plugin that can provision volume snapshot is needed. StorageClass and VolumeSnapshotClass are needed to deploy as well.

For example, you can use rook-ceph rbd plugin. In this case, make sure you are using ceph-csi version above 2.0.0. A bugfix for volume snapshot is applied on this version.


Install Kubevirt-Image-Service

# Download kubevirt-image-service project 
$ git clone https://github.com/tmax-cloud/kubevirt-image-service.git

# Select release version that you want to install
$ cd kubevirt-image-service/
$ git checkout v1.0.0

# Deploy CRD
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_virtualmachineimages_crd.yaml
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_virtualmachinevolumes_crd.yaml
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_virtualmachinevolumeexports_crd.yaml

# Deploy operator
$ kubectl apply -f deploy/namespace.yaml
$ kubectl apply -f deploy/role.yaml
$ kubectl apply -f deploy/role_binding.yaml
$ kubectl apply -f deploy/service_account.yaml
$ kubectl apply -f deploy/operator.yaml

# Check operator status
$ kubectl get deploy -n kis 
NAME                     READY   UP-TO-DATE   AVAILABLE   AGE
kubevirt-image-service   3/3     3            3           23s

Use Kubevirt-Image-Service

Import image

1. Import image from HTTP source

vmim is the shortname for VirtualMachineImage.

# Deploy image CR. See the following yaml file for more information about each CR fields
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_v1alpha1_virtualmachineimage_http_cr.yaml

# Wait until image state is ready to use
$ kubectl get vmim
NAME       STATE
myubuntu   Available

2. Import image from hostpath source

# Create a qcow2 image file with the name `disk.img` in the desired path (/mnt/data).

# Deploy host path image CR using the path that created the qcow2 file.
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_v1alpha1_virtualmachineimage_hostpath_cr.yaml

# Wait until image state is ready to use
$ kubectl get vmim
NAME       STATE
localvmim  Available

# When the status of vmim becomes Availalbe, user can delete the qcow2 file.

Create volume from image

vmv is the shortname for VirtualMachineVolume.

# Deploy volume CR
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_v1alpha1_virtualmachinevolume_cr.yaml

# Wait until volume state is ready to use
$ kubectl get vmv
NAME         STATE
myrootdisk   Available

Use created volume for VM

On VM yaml file, add disks and volumes section with PVC and disk information

apiVersion: kubevirt.io/v1alpha3
kind: VirtualMachine
metadata:
  name: vm
spec:
  template:
    spec:
      domain:
        devices:
          disks:
          - name: disk0
            disk:
              bus: virtio
      volumes:
      - name: disk0
        persistentVolumeClaim:
          claimName: myubuntu-pvc # name of pvc created by volume, {$VmvName}-vmv-pvc

Export volume to local destination

vmve is the shortname for VirtualMachineExport.

# Deploy export CR
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_v1alpha1_virtualmachinevolumeexport_cr.yaml

# Wait until export is completed
$ kubectl get vmve
NAME          STATE
disk-export   Completed

# Check exported pod
$ kubectl get pod
NAME                              READY   STATUS    RESTARTS   AGE
disk-export-exporter-local        1/1     Running   6          67m

# Copy exported volume to local path
# export/disk.img is in qcow2 format
# kubectl cp {vmve exporter pod name}:export/disk.img {local path to download}
$ kubectl cp disk-export-exporter-local:export/disk.img localpath.img

Export volume to external object storage

# Create k8s secret with your accessKeyId and secretAccessKey of the external endpoint
$ kubectl apply -f deploy/example/endpoint-secret.yaml

# Deploy export CR
# Your s3 endpoint must be written in virtual hosted style, https://bucket-name.endpoint/object-key-name
# You need to specify secretRef as well with the secret you created in the previous step
$ kubectl apply -f deploy/crds/hypercloud.tmaxanc.com_v1alpha1_virtualmachinevolumeexport_s3_cr.yaml

# Wait until export is completed
$ kubectl get vmve
NAME          STATE
s3-export   Completed