Ansible KubeVirt Modules

Ansible KubeVirt modules enable the automation of management of the following Kubernetes cluster object types:

• Virtual Machines (also VM templates and VM presets),
• VM Replica Sets,
• and Persistent Volume Claims (including Containerized Data Importer functionality).

Since the release of Ansible 2.8, the modules, the inventory plugin and relevant unit tests are part of the upstream Ansible git repository, while this repository contains only the integration tests and example playbooks.

Table of Contents

• Quickstart
• Requirements
• Source Code
• Testing
  • Automatic testing
  • Manual testing

Quickstart

For a quick introduction, please see the following kubevirt.io blog posts:

• KubeVirt with Ansible, part 1 – Introduction
• KubeVirt with Ansible, part 2

Requirements

• Ansible >= 2.8
  • pip3 --user install ansible
• openshift-restclient-python >= 0.8.2
  • pip3 --user install openshift

Source Code

• modules
  • module_utils
• inventory plugin
• unit tests
  • fixtures

Testing

There are two target of tests that can be found here:

• module tests in tests/playbooks
• role tests in tests/roles

To run a full complement of tests for a given target please use the relevant all.yml playbook.

Automatic testing

Unit tests

Upstream ansible repository contains unit tests covering the kubevirt modules.

Integration tests

Module tests (tests/playbooks/all.yml are run against actual clusters with both KubeVirt and CDI deployed, on top of:

• TravisCI (ubuntu vms supporting only minikube; no kvm acceleration for KubeVirt vms)
• oVirt Jenkins (physical servers that run any cluster kubevirtci supports)

Module tests are run using:

• most recently released ansible (whatever one gets with pip install ansible)
• ansible stable branch(es)
• ansible devel branch

Role tests (tests/roles/all.yml) are only run on TravisCI using the devel branch.

To detect regressions early, Travis runs all the tests every 24 hours against a fresh clone of ansible.git and emails kubevirt module developers if tests fail.

Manual testing

1. Clone this repository to a machine where you can oc login to your cluster:

   $ git clone https://github.com/kubevirt/ansible-kubevirt-modules.git
   $ cd ./ansible-kubevirt-modules
   

2. (Optional) Configure a virtual environment to isolate dependencies:

   $ python3 -m venv env
   $ source env/bin/activate
   

3. Install dependencies:

   $ pip install openshift
   

   If you skipped the previous step, you might need to prepend that command with sudo.

4. Install ansible (in one of the many ways):

   • Install the latest released version:

     $ pip install ansible
     

     Again, sudo might be required here.

   • Build RPM from the devel branch:

     $ git clone https://github.com/ansible/ansible.git
     $ cd ./ansible
     $ make rpm
     $ sudo rpm -Uvh ./rpm-build/ansible-*.noarch.rpm
     

   • Check out PRs locally

5. Run the tests:

   $ ansible-playbook tests/playbooks/all.yml
   

   Note: The playbook examples include cloud-init configuration to be able to access the created VMIs.

   1. For using SSH do as follows:

      $ kubectl get all
      NAME                             READY     STATUS    RESTARTS   AGE
      po/virt-launcher-bbecker-jw5kk   1/1       Running   0          22m
      
      $ kubectl expose pod virt-launcher-bbecker-jw5kk --port=27017 --target-port=22 --name=vmservice
      $ kubectl get svc vmservice
      NAME        TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)     AGE
      vmservice   ClusterIP   172.30.133.9   <none>        27017/TCP   19m
      
      $ ssh -i tests/test_rsa -p 27017 [email protected]
      

      It might take a while for the VM to come up before SSH can be used.

   2. For using virtctl:

      $ virtctl console <vmi_name>
      

      Or

      $ virtctl vnc <vmi_name>
      

      Use the username kubevirt and the password kubevirt.

6. (Optional) Leave the virtual environment and remove it:

   $ deactivate
   $ rm -rf env/
   

Notes on kubevirt_cdi_upload module

To upload an image from localhost by using the kubevirt_cdi_upload module, your system needs to be able to connect to the cdi upload proxy pod. This can be achieved by either:

1. Exposing the cdi-uploadproxy Service from the cdi namespace, or

2. Using kubectl port-forward to set up a temporary port forwarding through the Kubernetes API server: kubectl port-forward -n cdi service/cdi-uploadproxy 9443:443

Notes on the k8s_facts module

The following command will collect facts about the existing VM(s), if there are any, and print out a JSON document based on KubeVirt VM spec:

$ ansible-playbook examples/playbooks/k8s_facts_vm.yml

Notes on the KubeVirt inventory plugin

Inventory plugins allow users to point at data sources to compile the inventory of hosts that Ansible uses to target tasks, either via the -i /path/to/file and/or -i 'host1, host2' command line parameters or from other configuration sources.

Enabling the KubeVirt inventory plugin

To enable the KubeVirt plugin, add the following section in the tests/ansible.cfg file:

[inventory]
enable_plugins = kubevirt

Configuring the KubeVirt inventory plugin

Define the plugin configuration in tests/playbooks/plugin/kubevirt.yaml as follows:

plugin: kubevirt
connections:
  - namespaces:
      - default
    interface_name: default

In this example, the KubeVirt plugin will list all VMIs from the default namespace and use the default interface name.

Using the KubeVirt inventory plugin

To use the plugin in a playbook, run:

$ ansible-playbook -i kubevirt.yaml <playbook>

Note: The KubeVirt inventory plugin is designed to work with Multus. It can be used only for VMIs, which are connected to the bridge and display the IP address in the Status field. For VMIs exposed by Kubernetes services, please use the k8s Ansible module.