KubeVirt with Ansible, part 2

Part 1 contained a short introduction to basic VM management with Ansible’s kubevirt_vm module. This time we’ll paint a more complete picture of all the features on offer.

As before, examples found herein are also available as full working playbooks in our playbooks example repository. Additionally, each section of this post links to the corresponding module’s Ansible documentation page. Those pages always contain an Examples section, which the reader is encouraged to look through, as they have many more ways of using the modules than can reasonably fit here.

More VM management

Virtual machines managed by KubeVirt are highly customizable. Among the features accessible from Ansible, are:

  • various libvirt–level virtualized hardware tweaks (e.g. machine_type or cpu_model),
  • network interface configuration (interfaces), including multi–NIC utilizing the Multus CNI,
  • non–persistent VMs (ephemeral: yes),
  • direct DataVolumes support (datavolumes),
  • and OpenShift Templates support (template).

Further resources

VM Image Management with the Containerized Data Importer

The main functionality of the kubevirt_pvc module is to manage Persistent Volume Claims. The following snippet should seem familiar to anyone who dealt with PVCs before:

  name: pvc1
  namespace: default
  size: 100Mi
    - ReadWriteOnce

Running it inside a playbook will result in a new PVC named pvc1 with the access mode ReadWriteOnce and at least 100Mi of storage assigned.

The option dedicated to working with VM images is named cdi_source and lets one fill a PVC with data immediately upon creation. But before we get to the examples, the Containerized Data Importer needs to be properly deployed, which is as simple as running the following commands:

export CDI_VER=$(curl -s https://github.com/kubevirt/containerized-data-importer/releases/latest | grep -o "v[0-9]\.[0-9]*\.[0-9]*")
kubectl apply -f https://github.com/kubevirt/containerized-data-importer/releases/download/$CDI_VER/cdi-operator.yaml
kubectl apply -f https://github.com/kubevirt/containerized-data-importer/releases/download/$CDI_VER/cdi-cr.yaml

Once kubectl get pods -n cdi confirms all pods are ready, CDI is good to go.

The module can instruct CDI to fill the PVC with data from:

  • a remote HTTP(S) server (http:),
  • a container registry (registry:),
  • a local file (upload: yes), though this requires using kubevirt_cdi_upload for the actual upload step,
  • or nowhere (the blank: yes option).

Here’s a simple example:

name: pvc2
namespace: default
size: 100Mi
  - ReadWriteOnce
wait: yes
    url: https://download.cirros-cloud.net/0.4.0/cirros-0.4.0-x86_64-disk.img

info Please notice the wait: yes parameter. The module will only exit after CDI has completed transferring its data.

Let’s see this in action:

[mmazur@klapek part2]$ ansible-playbook pvc_cdi.yaml

TASK [Create pvc and fetch data] **********************************************************************************
changed: [localhost]

PLAY RECAP ********************************************************************************************************
localhost                  : ok=2    changed=1    unreachable=0    failed=0    skipped=0    rescued=0    ignored=0

[mmazur@klapek part2]$ kubectl get pvc
pvc2      Bound     local-pv-6b6380e2   37Gi       RWO            local          71s
[mmazur@klapek part2]$ kubectl get pvc/pvc2 -o yaml|grep cdi
    cdi.kubevirt.io/storage.import.endpoint: https://download.cirros-cloud.net/0.4.0/cirros-0.4.0-x86_64-disk.img
    cdi.kubevirt.io/storage.import.importPodName: importer-pvc2-gvn5c
    cdi.kubevirt.io/storage.import.source: http
    cdi.kubevirt.io/storage.pod.phase: Succeeded

Everything worked as expected.

Further resources

Inventory plugin

The default way of using Ansible is to iterate over a list of hosts and perform operations on each one. Listing KubeVirt VMs can be done using the KubeVirt inventory plugin. It needs a bit of setting up before it can be used.

First, enable the plugin in ansible.cfg:

enable_plugins = kubevirt

Then configure the plugin using a file named kubevirt.yml or kubevirt.yaml:

plugin: kubevirt
  - namespaces:
      - default
    network_name: default

And now let’s see if it worked and there’s a VM running in the default namespace (as represented by the namespace_default inventory group):

[mmazur@klapek part2]$ ansible -i kubevirt.yaml namespace_default --list-hosts
 [WARNING]: provided hosts list is empty, only localhost is available. Note that the implicit localhost does not
match 'all'

  hosts (0):

Right, we don’t have any VMs running. Let’s go back to part 1, create vm1, make sure it’s runing and then try again:

[mmazur@klapek part2]$ ansible-playbook  ../part1/02_vm1.yaml
PLAY RECAP ********************************************************************************************************
localhost                  : ok=2    changed=1    unreachable=0    failed=0    skipped=0    rescued=0    ignored=0

[mmazur@klapek part2]$ ansible-playbook  ../part1/01_vm1_running.yaml
PLAY RECAP ********************************************************************************************************
localhost                  : ok=2    changed=1    unreachable=0    failed=0    skipped=0    rescued=0    ignored=0

[mmazur@klapek part2]$ ansible -i kubevirt.yaml namespace_default --list-hosts
  hosts (1):


Further resources


Lastly, for the sake of brevity, a quick mention of the remaining modules:

  • kubevirt_presets allows setting up VM presets to be used by deployed VMs,
  • kubevirt_template brings in a generic templating mechanism, when running on top of OpenShift or OKD,
  • and kubevirt_rs lets one configure KubeVirt’s own ReplicaSets for running multiple instances of a specified virtual machine.