Description of problem:
During ocp-3.5 installation, although two nodes had the label matching router selector, but it still got "replicas": "1" when set_fact for router replicas.


Version-Release number of selected component (if applicable):
openshift-ansible-3.5.32-1.git.0.42cf266.el7.noarch.rpm

How reproducible:
Always

Steps to Reproduce:
1.Set openshift_hosted_router_selector in ansible inventory file and also have 2 nodes with corresponding node label specified.

[OSEv3:vars]
...

openshift_hosted_router_selector="role=node,router=enabled"


[nodes]
...

qe-gpei-ha-node-registry-router-1.0315-wqq.qe.rhcloud.com  openshift_hostname=qe-gpei-ha-node-registry-router-1 openshift_node_labels="{'role': 'node','registry': 'enabled','router': 'enabled'}"

qe-gpei-ha-node-registry-router-2.0315-wqq.qe.rhcloud.com  openshift_hostname=qe-gpei-ha-node-registry-router-2 openshift_node_labels="{'role': 'node','registry': 'enabled','router': 'enabled'}"



Actual results:
After installation, log into master to have a check
[root@qe-gpei-ha-master-etcd-1 ~]# oc get node --selector=role=node,router=enabled  -n default
NAME                                STATUS    AGE
qe-gpei-ha-node-registry-router-1   Ready     50m
qe-gpei-ha-node-registry-router-2   Ready     50m

[root@qe-gpei-ha-master-etcd-1 ~]# oc get pod |grep router
router-1-p1kgp             1/1       Running   0          45m

[root@qe-gpei-ha-master-etcd-1 ~]# oc get rc
NAME                 DESIRED   CURRENT   READY     AGE
docker-registry-1    2         2         2         30m
registry-console-1   1         1         1         30m
router-1             1         1         1         31m


Expected results:
router should have replicas=2

Additional info: