For evacuation, ignore if task_state is not None

ignore instance task state and continue with vm evacutaion

Closes-Bug: #1978983
Change-Id: I5540df6c7497956219c06cff6f15b51c2c8bc29d

doc/source/admin/evacuate.rst

@@ -97,3 +97,17 @@ instances up and running.
    using a pattern you might want to use the ``--strict`` flag which got introduced
    in version 10.2.0 to make sure nova matches the ``FAILED_HOST``
    exactly.
+
+.. note::
+   .. code-block:: bash
+
+      +------+--------+--------------+
+      | Name | Status | Task State   |
+      +------+--------+--------------+
+      | vm_1 | ACTIVE | powering-off |
+      +------------------------------+
+
+   If the instance task state is not None, evacuation will be possible. However,
+   depending on the ongoing operation, there may be clean up required in other
+   services which the instance was using, such as neutron, cinder, glance, or
+   the storage backend.

nova/compute/api.py

@@ -5476,7 +5476,7 @@ class API:
     @reject_vtpm_instances(instance_actions.EVACUATE)
     @block_accelerators(until_service=SUPPORT_ACCELERATOR_SERVICE_FOR_REBUILD)
     @check_instance_state(vm_state=[vm_states.ACTIVE, vm_states.STOPPED,
-                                    vm_states.ERROR])
+                                    vm_states.ERROR], task_state=None)
     def evacuate(self, context, instance, host, on_shared_storage,
                  admin_password=None, force=None):
         """Running evacuate to target host.
@@ -5503,7 +5503,7 @@ class API:
             context, instance.uuid)
 
         instance.task_state = task_states.REBUILDING
-        instance.save(expected_task_state=[None])
+        instance.save(expected_task_state=None)
         self._record_action_start(context, instance, instance_actions.EVACUATE)
 
         # NOTE(danms): Create this as a tombstone for the source compute

nova/tests/functional/regressions/test_bug_1978983.py

@@ -13,10 +13,8 @@
 # License for the specific language governing permissions and limitations
 # under the License.
 
-
 from nova import test
 from nova.tests import fixtures as nova_fixtures
-from nova.tests.functional.api import client
 from nova.tests.functional import fixtures as func_fixtures
 from nova.tests.functional import integrated_helpers
 
@@ -44,6 +42,7 @@ class EvacuateServerWithTaskState(
         api_fixture = self.useFixture(nova_fixtures.OSAPIFixture(
             api_version='v2.1'))
         self.api = api_fixture.admin_api
+        self.api.microversion = 'latest'
 
         self.src = self._start_compute(host='host1')
         self.dest = self._start_compute(host='host2')
@@ -53,26 +52,20 @@ class EvacuateServerWithTaskState(
         """
         server = self._create_server(networks=[])
 
-        self.api.microversion = 'latest'
         server = self._wait_for_state_change(server, 'ACTIVE')
-        self.assertEqual('host1', server['OS-EXT-SRV-ATTR:host'])
+        self.assertEqual(self.src.host, server['OS-EXT-SRV-ATTR:host'])
 
         # stop host1 compute service
         self.src.stop()
+        self.api.put_service_force_down(self.src.service_ref.uuid, True)
 
         # poweroff instance
         self._stop_server(server, wait_for_stop=False)
         server = self._wait_for_server_parameter(
             server, {'OS-EXT-STS:task_state': 'powering-off'})
 
-        # FIXME(auniyal): As compute service is down in source node
-        # instance is stuck at powering-off, evacuation fails with
-        # msg: Cannot 'evacuate' instance <instance-id> while it is in
-        # task_state powering-off (HTTP 409)
-
-        ex = self.assertRaises(
-            client.OpenStackApiException,
-            self._evacuate_server,
-            server,
-            expected_host=self.dest.host)
-        self.assertEqual(409, ex.response.status_code)
+        # evacuate instance
+        server = self._evacuate_server(
+            server, expected_host=self.dest.host
+        )
+        self.assertEqual(self.dest.host, server['OS-EXT-SRV-ATTR:host'])

releasenotes/notes/ignore-instance-task-state-for-evacuation-e000f141d0153638.yaml

@@ -0,0 +1,11 @@
+---
+fixes:
+  - |
+    If compute service is down in source node and user try to stop
+    instance, instance gets stuck at powering-off, hence evacuation fails with
+    msg: Cannot 'evacuate' instance <instance-id> while it is in
+    task_state powering-off.
+    It is now possible for evacuation to ignore the vm task state.
+    For more details see: `bug 1978983`_
+
+    .. _`bug 1978983`: https://bugs.launchpad.net/nova/+bug/1978983