Add mariadb override to work around unclean shutdown

Currently on a lock/unlock the mariadb pod does not do a clean
shutdown.  Because of this, on startup it doesn't want to do an
automatic recovery.

To work around this, add an override that tells the chart to
always recover on the one mariadb pod that we have.

Once we move to a galera cluster, we'll need to change this.

Story: 2003909
Task: 27086
Signed-off-by: Chris Friesen <chris.friesen@windriver.com>

Change-Id: I78f5dff402ad90751901f510a2ace973b7783d7f

sysinv/sysinv/sysinv/sysinv/helm/mariadb.py

@@ -25,6 +25,9 @@ class MariadbHelm(openstack.OpenstackBaseHelm):
         return self.SUPPORTED_NAMESPACES
 
     def get_overrides(self, namespace=None):
+        # NOTE: there is an intentional typo in "force_recovey" to match
+        # a typo in the upstream helm chart.  This will eventually go away
+        # when we upgrade to the latest chart.
         overrides = {
             common.HELM_NS_OPENSTACK: {
                 'pod': {
@@ -33,6 +36,7 @@ class MariadbHelm(openstack.OpenstackBaseHelm):
                     }
                 },
                 'endpoints': self._get_endpoints_overrides(),
+                'force_recovey': 'mariadb-server-0',
             }
         }