Using vmware_guest_snapshot with case insensitive VM names

Archives Ansible Project
1

Hi,

I’m looking at switching to using the vmware_guest_snapshot module to take snapshots of VMs before performing package updates on them (I previously used another custom module where we changed case of hostnames to lowercase and compared to vm object name after also making that lowercase). Unfortunately, our VMs are not consistently named in vCenter - they can be lowercase, uppercase or even a mixture of both - but the module only supports finding VMs using the exact case. This makes it difficult when passing hostnames to a playbook, we always pass a lowercase and the hostnames are used as the VM name by my playbook.

I’ve thought about using a daily cache file created using the vmware_vm_facts module (we don’t want to have to query vCentre for these facts on each run as it contains hundreds of VMs), retrieving the machine UUID from that cache and using it to identify the VM to snapshot (rathre than its name). However, I’ve not been able to figure out how to select from my cache file (without actually making it all lowercase).

So to debug, the following just writes out the facts gathered by vmware_vm_facts and then the debug grabs the UUID by matching the inventory hostname (but obviously that only works if it matches the case sensitive hostname);

  • name: Write out facts to file
    copy: content=“{{ vm_facts.virtual_machines }}” dest=vm_facts.cache
    delegate_to: localhost
    run_once: ‘True’

  • debug:
    msg: “My UUID is {{ myvm[inventory_hostname][‘uuid’] }}”
    vars:
    myvm: “{{ lookup(‘file’, ‘vm_facts.cache’) }}”

I know I can work with this by making things all lower case as follows;

  • debug:
    msg: “My UUID is {{ myvm[inventory_hostname|lower][‘uuid’] }}”
    vars:
    myvm: “{{ lookup(‘file’, ‘vm_facts.cache’) | lower }}”

Sample cache file (ignore values - they’ve been changed);

{“SERVER1”: {“esxi_hostname”: “test_esx”, “vm_network”: {“00:00:00:00:00”: {“ipv4”: [“10.0.0.1”], “ipv6”: [“f----”]}}, “power_state”: “poweredOn”, “mac_address”: [“00:00:00:00:00”], “guest_fullname”: “CentOS 4/5 or later (64-bit)”, “ip_address”: “10.0.0.1”, “uuid”: “422342435-de04-5014-58d4-475ab4ffdc8a”}, “server2”: {“esxi_hostname”: “test_esx”, “vm_network”: {}, “power_state”: “poweredOn”, “mac_address”: [“00:00:00:00:00:00”], “guest_fullname”: “CentOS 4/5 or later (64-bit)”, “ip_address”: “”, “uuid”: “422344329-d4df-72c9-477b-bdd976e72be7”}}

But can anyone advise a better way of achieving this without having to change everything to lowercase (just in case UUID ever contains upper and lowercase characters)? Alternatively, a way to find the VM name from the cache file and use that to pass as the name to the module.

It would be nice if the vmware modules would allow case insensitive matches of VM names (the modules already cater for a situation where more than one VM has the same name by using the name_match parameter).

Ansible Version: 2.6.5

2

I’ve just found the UUID is avaible in ansible_product_uuid (or facter fact as ‘uuid’). So easy enough if we gather facts from the machine first.