Update docs to reflect requirements on port and image names.

deploy/README.md

@@ -69,7 +69,8 @@ This example uses a Deployment to run the controller container in the
 ```
 $ kubectl apply -f controller.yaml
 ```
-The ``image`` in the manifest MUST be ``varnish-ingress/controller``.
+The ``image`` in the manifest must specify the controller (named
+``varnish-ingress/controller`` above).
 
 It does *not* make sense to deploy more than one replica of the
 controller. If there are more controllers, all of them will connect to
@@ -149,7 +150,8 @@ There are some requirements on the configuration of the Varnish
 deployment that must be fulfilled in order for the Ingress to work
 properly:
 
-* The ``image`` must be specified as ``varnish-ingress/varnish``.
+* The ``image`` must specify the Varnish container (named
+  ``varnish-ingress/varnish`` above)
 * ``spec.template`` must specify the label ``app: varnish-ingress``.
   The controller recognizes Services with this label as Varnish
   deployments meant to implement Ingress:
@@ -266,6 +268,12 @@ The Service definition must fulfill some requirements:
 
   **TO DO**: The ``port.name`` is hardwired to ``varnishadm``.
 
+* A port with the name ``http`` MUST be specified, whose
+  ``targetPort`` matches the http port defined above (named ``http``
+  in the sample Deployment).
+
+  **TO DO**: The ``port.name`` is hardwired to ``http``.
+
 * The Service must be defined so that the cluster API will allow
   Endpoints to be listed when the container is not ready (since
   the Varnish instances are initialized in the not ready state).