ForceDelete Instance

If nova config 'reclaim_instance_interval', a instance will be
soft-deleted, that will cause a error, that volumes or other attached with the
instance will cann't be deleted for its 'in-use' status.
Now force_delete the instance, the error no longer occurs.

Closes-Bug: #1564265
Change-Id: I4d1dd33b23d5882481f4d2c06107c3fb8c08cb93

heat/engine/clients/os/nova.py

@@ -445,9 +445,12 @@ echo -e '%s\tALL=(ALL)\tNOPASSWD: ALL' >> /etc/sudoers
             return False
 
         status = self.get_status(server)
-        if status in ("DELETED", "SOFT_DELETED"):
+        if status == 'DELETED':
             return True
-        if status == 'ERROR':
+
+        if status == 'SOFT_DELETED':
+            self.client().servers.force_delete(server_id)
+        elif status == 'ERROR':
             fault = getattr(server, 'fault', {})
             message = fault.get('message', 'Unknown')
             code = fault.get('code')

heat/tests/openstack/nova/fakes.py

@@ -287,6 +287,8 @@ class FakeSessionClient(base_client.SessionClient):
             assert set(body[action].keys()) == set(['host',
                                                     'block_migration',
                                                     'disk_over_commit'])
+        elif action == 'forceDelete':
+            assert body is not None
         else:
             raise AssertionError("Unexpected server action: %s" % action)
         return (resp, _body)

releasenotes/notes/force-delete-nova-instance-6ed5d7fbd5b6f5fe.yaml

@@ -0,0 +1,9 @@
+---
+fixes:
+  - Force delete the nova instance.
+    If a resource is related with a nova instance which
+    is in 'SOFT_DELETED' status, the resource can't be
+    deleted, when nova config 'reclaim_instance_interval'.
+    so, force-delete the nova instance, and then all the
+    resources are related with the instance would be
+    processed properly.