Move ansible_ssh_host to ansible_host

We've moved all roles/tasks to use ansible_host instead of
ansible_ssh_host, since this is the new approach for Ansible.

There were a few left-overs which this PR fixes:
* Documentation in sample swift-remote.yml file
* Print hosts (IP) for the inventory-manage.py script lib.
* galera-cluster-rolling-restart - this will default to ansible_ssh_host
if ansible_host is not specified, although it should be if the inventory
has run.

Change-Id: I45e47e6d31966598c60c938861630e182d31efe5

etc/openstack_deploy/conf.d/swift-remote.yml.sample

@@ -2,7 +2,7 @@
 # For remote swift hosts you will only specify the IP address
 # NB. The ip address that must be specified is the IP address that swift
 # hosts will be able to replicate on and not necessarily the
-# ansible_ssh_host IP from the remote ansible install.
+# ansible_ssh IP from the remote ansible install.
 swift-remote_hosts:
   remote-swift1:
     ip: 172.29.237.1

lib/manage.py

@@ -249,7 +249,7 @@ def print_inventory(inventory, sort_key):
         'component',
         'physical_host',
         'tunnel_address',
-        'ansible_ssh_host',
+        'ansible_host',
         'container_types'
     ]
     table = prettytable.PrettyTable(required_list)

scripts/upgrade-utilities/playbooks/galera-cluster-rolling-restart.yml

@@ -44,7 +44,7 @@
       local_action:
         module: wait_for
         port: "3306"
-        host: "{{ ansible_ssh_host | default(inventory_hostname) }}"
+        host: "{{ ansible_host | default(ansible_ssh_host | default(inventory_hostname)) }}"
       retries: 10
       delay: 10