Getting `host_vars` From The Inventory For A Host That Doesn't Yet Exist (But Which Is Listed In The Inventory File)

Get Help
1

Hi All,

I’m looking for some advice / pointers / ideas on the following situation:

Let me outline what I’ve got / what I’m thinking and then what I’m trying to achieve with that thinking.

My Inventory file has a number of groups, sub-groups, and (obviously) hosts.

I use the FQDN as the inverntory_hostname.

Each group, sub-group, and host has a number of group_vars or host_vars associated with it, and the whole Inventory file is structured in such a way that:

  • Only the unique host_vars for a host appear in that host’s Var file
  • Only those group_vars that are common to a sub-group’s host(s) (ie unique to that sub-group) appear in that sub-group’s Var file
  • Only only those group_vars that are common to a group’s sub-group(s) and/or host(s) (ie unique to that group) appear in that group’s Var file

The idea here is that (as I understand things - and I could be wrong) the group-vars “cascade down” to the individual hosts and are merged into the set of host_vars.

One of the set of host_vars/group_vars for (VM) hosts are the specifications for that host ie Number of CPUs, RAM, VLAN, IPv4 & IPv6 Addresses, MAC Address(s), etc. I have a “Golden Image” VM set up with the most commonly used (ie default) set of these settings which I currently clone and (as required) modify when I need a new VM - which, to be honest, isn’t all that often.

What I’m thinking is that I can use Ansible to perform the “clone & modify” by providing the playbook with the hostname (or FQDN) of the new soon-to-exist VM and (somehow) how Ansible extract the relevant CPUs, RAM, etc, variables from the Inventory file, even though the actual host won’t exist until the end of the playbook run.

The command to start the playbook would be something like: ansible-playbook -i localhost clone_vm.yml, and inside the play there is a vars_prompt: (say vm_name) asking for the new VM’s hostname (or FQDN).

Is something like this viable (I’d like to think that is would be, as I already have a bash script which does all this - although I’ve got to provide all the values manually when I call the script) and if so, how do I take the vars_prompt: vm_name and use that as a key to lookup the other host_vars specific to the soon-to-be new host (and, of course, the corresponding group_vars in the new host’s sub-group(s) and group)?

Thanks in advance for the help

Cheers

Dulux-Oz

2

Hello @Dulux-Oz

As I understand, you want to use already defined facts/variables in an Ansible group for a not-existing VM. That is possible using the add_host module. Once you add the vm to the groups which has the required facts and variables, you can use any of them.

Here is an example clone_vm.yml

- hosts: localhost
  vars_prompt:
    - name: vm_name
      prompt: What's VM hostname ?
    - name: vm_groups
      prompt: Ansible groups to add the new host # <- list of groups
    # Any other prompt you need

  pre_tasks: # run before any roles, if you have any
    - name: Add host to in-memory inventory
      add_host:
        name: "{{ vm_name }}" # defined in vars_prompt
        groups: "{{ vm_groups }}" # defined in vars_prompt

  tasks:
    - name: Creating "{{ vm_name }}"
      [... your tasks ...]

You can also define an extra-vars.yml file intead of prompting all the variables and call it from the ansible command: ansible-playbook -i [inventory_file] clone_vm.yml -e @extra-vars.yml