You need to set --images='registry.ops.openshift.com/openshift3/ose-${component}:${version}'

oadm router / registry / ipfailover have never been smart enough to query the API for image name expansion so you have to pass it in to the command line.

# oadm ipfailover ipf-har --replicas=1 --watch-port 80 --selector="infra=ha-router" --virtual-ips="172.18.1.153" --service-account=admin --images='registry.ops.openshift.com/openshift3/ose-${component}:${version}' -o yaml | grep 'image:'

          image: registry.ops.openshift.com/openshift3/ose-keepalived-ipfailover:v3.6.170

If --images='registry.ops.openshift.com/openshift3/ose-${component}:${version}' is mandatory option to deploy ipfailover pod, we may need update this in our documents in https://docs.openshift.com/container-platform/3.5/admin_guide/high_availability.html and documents for 3.6 release


## Test passed with --image option

[root@ip-172-18-3-177 ~]# oadm ipfailover ipf-har --replicas=1 --watch-port=80 --selector="infra=ha-router" --virtual-ips="172.18.1.153" --service-account=harp --images='registry.ops.openshift.com/openshift3/ose-${component}:${version}'
--> Creating IP failover ipf-har ...
    error: serviceaccounts "harp" already exists
    deploymentconfig "ipf-har" created
--> Failed
[root@ip-172-18-3-177 ~]# oc get pods
NAME               READY     STATUS              RESTARTS   AGE
ipf-har-1-deploy   1/1       Running             0          7s
ipf-har-1-gvmzs    0/1       ContainerCreating   0          1s
[root@ip-172-18-3-177 ~]# oc get pods
NAME              READY     STATUS    RESTARTS   AGE
ipf-har-1-gvmzs   1/1       Running   0          27s
[root@ip-172-18-3-177 ~]#