Description of problem:
User should be able to upgrade from 3.4 to 3.5 Metrics, this should do what deployer used to do when set MODE=refresh, pv and route should be kept, but pods are updated with new version of images. As we discussed in the card, this should be done by just run the playbook.
However, after run the playbook, previous version is still there.

Version-Release number of selected component (if applicable):
OCP 3.5
openshift-ansible master branch

How reproducible:
always

Steps to Reproduce:
1. Deploy 3.4 Metrics and check

oc project openshift-infra

oc create -f - <<API
apiVersion: v1
kind: ServiceAccount
metadata:
  name: metrics-deployer
secrets:
- name: metrics-deployer
API

oadm policy add-role-to-user edit system:serviceaccount:openshift-infra:metrics-deployer

oc secrets new metrics-deployer nothing=/dev/null

oadm policy add-cluster-role-to-user cluster-reader system:serviceaccount:openshift-infra:heapster

oadm policy add-role-to-user view system:serviceaccount:openshift-infra:hawkular -n openshift-infra

oc new-app -f metrics.yaml --as=system:serviceaccount:openshift-infra:metrics-deployer \
-p IMAGE_PREFIX=$PREFIX \
-p IMAGE_VERSION=3.4.1 \
-p HAWKULAR_METRICS_HOSTNAME=hawkular-metrics.$SUBDOMAIN \
-p MODE=deploy \
-p USE_PERSISTENT_STORAGE=true \
-p MASTER_URL=$MASTERURL \
-p DYNAMICALLY_PROVISION_STORAGE=true \
-p CASSANDRA_NODES=1 \
-p CASSANDRA_PV_SIZE=5Gi \
-p USER_WRITE_ACCESS=false

# oc get pod
NAME                         READY     STATUS      RESTARTS   AGE
hawkular-cassandra-1-z5b7w   1/1       Running     0          1h
hawkular-metrics-n7jkp       1/1       Running     0          1h
heapster-jnkv6               1/1       Running     0          1h
metrics-deployer-vvzbl       0/1       Completed   0          1h

# oc get pvc
NAME                  STATUS    VOLUME                                     CAPACITY   ACCESSMODES   AGE
metrics-cassandra-1   Bound     pvc-550a6728-edca-11e6-bb92-0e9aebfd1b9e   5Gi        RWO           1h

2. prepare inventory file

[oo_first_master]
$MASTER ansible_user=root ansible_ssh_user=root ansible_ssh_private_key_file="~/.ssh/libra.pem" openshift_public_hostname=$MASTER

[oo_first_master:vars]
deployment_type=openshift-enterprise
openshift_release=v3.5.0

openshift_metrics_install_metrics=true

openshift_metrics_hawkular_hostname=hawkular-metrics.$SUBDOMAIN
openshift_metrics_project=openshift-infra

openshift_metrics_image_prefix=registry.ops.openshift.com/openshift3/
openshift_metrics_image_version=3.5.0

openshift_metrics_cassandra_storage_type=dynamic
openshift_metrics_cassandra_pv_size=5Gi

3. run playbook
git clone https://github.com/openshift/openshift-ansible
ansible-playbook  -vvv  -i  ~/inventory  playbooks/common/openshift-cluster/openshift_metrics.yml

Actual results:
# oc get pod
NAME                         READY     STATUS      RESTARTS   AGE
hawkular-cassandra-1-z5b7w   1/1       Running     0          2h
hawkular-metrics-n7jkp       1/1       Running     0          2h
heapster-jnkv6               1/1       Running     0          2h
metrics-deployer-vvzbl       0/1       Completed   0          2h


Expected results:
Previous Metrics should be updated, metrics data and route should still be there

Additional info:
Ansible log is attached.