ensure the volume pool can be marked autostart in all cases

In some cases after rerunning the workflow it is possible to
get into a state where the oooq pool is running but not
autostarted.  In this case the pool_check will pass and pool
autostart will fail.  This should provide the steps required
for autostart to work in either case.

Closes-Bug: #1657232

Change-Id: Ieba0caa40290c9e03740cef337c6a3f3fa04db30

roles/libvirt/setup/overcloud/tasks/main.yml

@@ -30,6 +30,28 @@
     name: "{{ libvirt_volume_pool }}"
     uri: "{{ libvirt_uri }}"
 
+# In some cases the pool_check can pass and the pool xml config is absent
+# In this case it is required to dump the xml and redefine the pool
+# I believe this is the case when {{ working_dir }}/volume_pool.xml is not present.
+# https://bugs.launchpad.net/tripleo-quickstart/+bug/1657232
+
+- name: check for the existence of {{ working_dir }}/volume_pool.xml
+  stat:
+    path: "{{ working_dir }}/volume_pool.xml"
+  register: volume_pool
+
+- name: dump the running virsh pool's xml file to the system
+  shell: "virsh pool-dumpxml {{ libvirt_volume_pool }} > {{ working_dir }}/volume_pool.xml"
+  when: volume_pool.stat.exists == False
+  environment:
+    LIBVIRT_DEFAULT_URI: "{{ libvirt_uri }}"
+
+- name: re-define the volume pool
+  command: "virsh pool-define {{ working_dir }}/volume_pool.xml"
+  when: volume_pool.stat.exists == False
+  environment:
+    LIBVIRT_DEFAULT_URI: "{{ libvirt_uri }}"
+
 - name: Mark volume pool for autostart
   virt_pool:
     name: "{{ libvirt_volume_pool }}"