Description of problem:
Upgrade ocp with dedicated etcd from v3.6 to v3.7, upgrade failed at task [Stop atomic-openshift-master-controllers] due to wrong hosts group.

TASK [Stop atomic-openshift-master-controllers] *****************************************************************************************************************************
fatal: [x.x.x.x]: FAILED! => {"changed": false, "failed": true, "msg": "Could not find the requested service atomic-openshift-master-controllers: host"}
	to retry, use: --limit @/usr/share/ansible/openshift-ansible/playbooks/byo/openshift-cluster/upgrades/v3_7/upgrade.retry

[OSEv3:children]
masters
nodes
etcd
...
[masters]
A
..
[etcd]
B

- name: Cycle all controller services to force new leader election mode
  hosts: oo_etcd_to_config
  gather_facts: no
  tasks:
  - name: Stop {{ openshift.common.service_type }}-master-controllers
    systemd:
      name: "{{ openshift.common.service_type }}-master-controllers"
      state: stopped
  - name: Start {{ openshift.common.service_type }}-master-controllers
    systemd:
      name: "{{ openshift.common.service_type }}-master-controllers"
      state: started


Version-Release number of the following components:
openshift-ansible-3.7.0-0.167.0.git.0.0e34535.el7.noarch
ansible-2.4.1.0-0.1.beta2.el7.noarch

How reproducible:
always

Steps to Reproduce:
1. Install ocp 3.6 with dedicated etcd
2. Upgrade 3.6 to 3.7
3.

Actual results:
Upgrade failed.

Expected results:
Upgrade succeed.

Additional info:
Please attach logs from ansible-playbook with the -vvv flag