vmware_portgroup: Can't remove a portgroup

[mariolenz]

SUMMARY

While working on the CI to make it compatible with ansible-core 2.19 (#2388) I've seen the following:

2025-04-26 11:21:23.156655 | controller | TASK [prepare_vmware_tests : Remove any potential existing "VM Network" on vSwitch0] ***
2025-04-26 11:21:23.156665 | controller | task path: /home/zuul/.ansible/collections/ansible_collections/community/vmware/tests/integration/targets/prepare_vmware_tests/tasks/setup_attach_hosts.yml:34
2025-04-26 11:21:23.156685 | controller | Using module file /home/zuul/.ansible/collections/ansible_collections/community/vmware/plugins/modules/vmware_portgroup.py
2025-04-26 11:21:23.840710 | controller | Pipelining is enabled.
2025-04-26 11:21:23.840736 | controller | <testhost> ESTABLISH LOCAL CONNECTION FOR USER: zuul
2025-04-26 11:21:23.840742 | controller | <testhost> EXEC /bin/sh -c '/home/zuul/venv/bin/python && sleep 0'
2025-04-26 11:21:23.840747 | controller | failed: [testhost] (item=esxi1.test) => {
2025-04-26 11:21:23.840751 | controller |     "ansible_loop_var": "item",
2025-04-26 11:21:23.840755 | controller |     "changed": false,
2025-04-26 11:21:23.840759 | controller |     "invocation": {
2025-04-26 11:21:23.840768 | controller |         "module_args": {
2025-04-26 11:21:23.840772 | controller |             "cluster_name": null,
2025-04-26 11:21:23.840776 | controller |             "esxi_hostname": "esxi1.test",
2025-04-26 11:21:23.840781 | controller |             "hostname": "vcenter.test",
2025-04-26 11:21:23.840785 | controller |             "hosts": [
2025-04-26 11:21:23.840789 | controller |                 "esxi1.test"
2025-04-26 11:21:23.840793 | controller |             ],
2025-04-26 11:21:23.840797 | controller |             "password": "VALUE_SPECIFIED_IN_NO_LOG_PARAMETER",
2025-04-26 11:21:23.840801 | controller |             "port": 443,
2025-04-26 11:21:23.840805 | controller |             "portgroup": "VM Network",
2025-04-26 11:21:23.840810 | controller |             "proxy_host": null,
2025-04-26 11:21:23.840814 | controller |             "proxy_port": null,
2025-04-26 11:21:23.840818 | controller |             "security": null,
2025-04-26 11:21:23.840822 | controller |             "state": "absent",
2025-04-26 11:21:23.840826 | controller |             "switch": "vSwitch0",
2025-04-26 11:21:23.840830 | controller |             "teaming": null,
2025-04-26 11:21:23.840837 | controller |             "traffic_shaping": null,
2025-04-26 11:21:23.840841 | controller |             "username": "administrator@vsphere.local",
2025-04-26 11:21:23.840845 | controller |             "validate_certs": false,
2025-04-26 11:21:23.840849 | controller |             "vlan_id": 0
2025-04-26 11:21:23.840853 | controller |         }
2025-04-26 11:21:23.840857 | controller |     },
2025-04-26 11:21:23.840862 | controller |     "item": "esxi1.test",
2025-04-26 11:21:23.840866 | controller |     "msg": "The portgroup already exists on vSwitch 'isolated_vSwitch'"
2025-04-26 11:21:23.840870 | controller | }
2025-04-26 11:21:23.840875 | controller | ...ignoring

It looks like it's impossible to remove a portgroup because the module seems to fail if it exists. This doesn't make sense to me.

ISSUE TYPE

• Bug Report

COMPONENT NAME

vmware_portgroup

ADDITIONAL INFORMATION

I think find_portgroup_by_name should just find the portgroup, but not fail if it exist:

community.vmware/plugins/modules/vmware_portgroup.py

Lines 388 to 390 in ce75a5a

	if portgroup.spec.name == portgroup_name and portgroup.spec.vswitchName != vswitch_name:
	# portgroup names are unique; there can be only one portgroup with the same name per host
	self.module.fail_json(msg="The portgroup already exists on vSwitch '%s'" % portgroup.spec.vswitchName)

[Rogacz]

Task is looking for port group "VM Network" on switch "vSwitch0" but it finds it on vSwitch "isolated_vSwitch"
If it didn't fail here in next step it would try to remove it, and it would remove "VM Network" on "isolated_vSwitch", not what was requested.

Alternatively, find_portgroup_by_name should return none.

But then if this was a task with state=present it would try to create it, and it can't because port group name has to be unique on the host, so we need a fail here.

Function check_if_portgroup_exists requires additional state, maybe "absent_with_confilct" so it will work for requested state=absent and failed for state=present.

[mariolenz]

I'm sorry, but my priority ATM is to make the collection work with the upcoming ansible-core 2.19 release (#2388 and #2391).

It turns out that this is harder than I originally expected. Please bear wit me 🙏