Fix auditd restart handler

It is not possible to restart auditd with systemctl. Using the service
interface is required. There are chef cookbooks[1] with the same
workaround.

This patch also includes a `cache_valid_time` addition to test.yml to
unblock the gate.

[1] https://github.com/chef-cookbooks/auditd/pull/22/files

Change-Id: I1aa3faf88f5953c230693600fcbcb786d49a35e0

handlers/main.yml

@@ -14,10 +14,13 @@
 # limitations under the License.
 
 # Restarting services ########################################################
+#
+# NOTE(mhayden): It's not possible to use systemd to restart auditd on CentOS
+# since it's a special service. Using the old service scripts is required.
 - name: restart auditd
-  service:
-    name: auditd
-    state: restarted
+  command: service auditd restart
+  args:
+    warn: no
 
 - name: restart chrony
   service:

tests/test.yml

@@ -19,6 +19,7 @@
     - name: Ensure apt cache is updated before testing
       apt:
         update_cache: yes
+        cache_valid_time: "{{ cache_timeout }}"
       when: ansible_pkg_mgr == 'apt'
     - name: Ensure OpenStack CI image has a logrotate cron job
       file: