Using the ovirt_vms module with cloud_init

A quick blog today. I've been wrestling with the ovirt_vms module in Ansible 2.5. I wanted to deploy a RHEL VM from a template on a RHV 4.1 cluster and use cloud-init to customize the VM.

The ovirt_vms module offers a cloud_init option that takes a custom_script parameter. According to the docs, you can feed this parameter your script like so:

- name: create VM
  ovirt_vms:
    name: myvm
    auth: "{{ ovirt_auth}}"
    [...] [etc...]
    wait: yes
    cloud_init:
      custom_script: |
        write_files:
        - path: /tmp/my-userdata.sh
          permissions: '0755'
          content: |                                                        
            #!/bin/bash
            echo "foo"

What this does not tell you, is that this userdata is never touched if you just drop it in the playbook like this. I mean, it makes sense, because you have to tell cloud-init to do something with the script, but the ovirt_vms documentation doesn't tell you how to do that. Therefore, the /tmp/my-userdata.sh script will be created, but it will not be executed.

After some experimenting, it turns out you have to pass the custom_script parameter a second statement after write_files. You need to pass it runcmd.

So if anyone is stumbling across this like I did, here's how to make your cloud-init work with ovirt_vms:

- name: create VM
  ovirt_vms:
    name: myvm
    auth: "{{ ovirt_auth}}"
    [...] [etc...]
    wait: yes
    cloud_init:
      custom_script: |
        write_files:
        - path: /tmp/my-userdata.sh
          permissions: '0755'
          content: |                                                        
            #!/bin/bash
            echo "foo"
        runcmd:
        - /tmp/my-userdata.sh

This may be obvious if you write cloud-init things all day, but I don't usually spend much time doing that :)