Set instance host field after resource claim

Set the 'host' field on the instance after the resource tracker
on the compute node has accepted the build.

The field is set after resources are confirmed to be available while the
COMPUTE_RESOURCES_SEMAPHORE is held.  The semaphore ensures the
resources usage values will be consistent even if the
update_available_resource periodic task audit runs.

bug 1060255

Change-Id: I92105ec14924960ac8ef7ca8c810783085314e10

nova/compute/manager.py

@@ -488,10 +488,6 @@ def _run_instance(self, context, request_spec,
                 limits = filter_properties.get('limits', {})
                 with self.resource_tracker.resource_claim(context, instance,
                         limits):
-                    # Resources are available to build this instance here,
-                    # mark it as belonging to this host:
-                    self._instance_update(context, instance['uuid'],
-                            host=self.host, launched_on=self.host)
 
                     block_device_info = self._prep_block_device(context,
                             instance)

nova/compute/resource_tracker.py

@@ -23,6 +23,7 @@
 from nova import db
 from nova import exception
 from nova import flags
+from nova import notifications
 from nova.openstack.common import cfg
 from nova.openstack.common import importutils
 from nova.openstack.common import jsonutils
@@ -152,8 +153,18 @@ def begin_resource_claim(self, context, instance_ref, limits=None,
                   failed.
         """
         if self.disabled:
+            # compute_driver doesn't support resource tracking, just
+            # set the 'host' field and continue the build:
+            instance_ref = self._set_instance_host(context,
+                                                   instance_ref['uuid'])
             return
 
+        # sanity check:
+        if instance_ref['host']:
+            LOG.warning(_("Host field should be not be set on the instance "
+                          "until resources have been claimed."),
+                          instance=instance_ref)
+
         if not limits:
             limits = {}
 
@@ -184,6 +195,8 @@ def begin_resource_claim(self, context, instance_ref, limits=None,
         if not self._can_claim_cpu(vcpus, vcpu_limit):
             return
 
+        instance_ref = self._set_instance_host(context, instance_ref['uuid'])
+
         # keep track of this claim until we know whether the compute operation
         # was successful/completed:
         claim = Claim(instance_ref, timeout)
@@ -196,6 +209,17 @@ def begin_resource_claim(self, context, instance_ref, limits=None,
         self._update(context, self.compute_node)
         return claim
 
+    def _set_instance_host(self, context, instance_uuid):
+        """Tag the instance as belonging to this host.  This should be done
+        while the COMPUTE_RESOURCES_SEMPAHORE is being held so the resource
+        claim will not be lost if the audit process starts.
+        """
+        values = {'host': self.host, 'launched_on': self.host}
+        (old_ref, instance_ref) = db.instance_update_and_get_original(context,
+                instance_uuid, values)
+        notifications.send_update(context, old_ref, instance_ref)
+        return instance_ref
+
     def _can_claim_memory(self, memory_mb, memory_mb_limit):
         """Test if memory needed for a claim can be safely allocated"""
         # Installed memory and usage info:

nova/scheduler/chance.py

@@ -68,7 +68,7 @@ def schedule_run_instance(self, context, request_spec,
                 host = self._schedule(context, 'compute', request_spec,
                                       filter_properties)
                 updated_instance = driver.instance_update_db(context,
-                        instance_uuid, host)
+                        instance_uuid)
                 self.compute_rpcapi.run_instance(context,
                         instance=updated_instance, host=host,
                         requested_networks=requested_networks,

nova/scheduler/driver.py

@@ -101,13 +101,13 @@ def cast_to_volume_host(context, host, method, **kwargs):
     LOG.debug(_("Casted '%(method)s' to volume '%(host)s'") % locals())
 
 
-def instance_update_db(context, instance_uuid, host):
-    '''Set the host and scheduled_at fields of an Instance.
+def instance_update_db(context, instance_uuid):
+    '''Clear the host and set the scheduled_at field of an Instance.
 
     :returns: An Instance with the updated fields set properly.
     '''
     now = timeutils.utcnow()
-    values = {'host': host, 'scheduled_at': now}
+    values = {'host': None, 'scheduled_at': now}
     return db.instance_update(context, instance_uuid, values)
 
 
@@ -116,7 +116,7 @@ def cast_to_compute_host(context, host, method, **kwargs):
 
     instance_uuid = kwargs.get('instance_uuid', None)
     if instance_uuid:
-        instance_update_db(context, instance_uuid, host)
+        instance_update_db(context, instance_uuid)
 
     rpc.cast(context,
              rpc.queue_get_for(context, 'compute', host),

nova/scheduler/filter_scheduler.py

@@ -142,8 +142,7 @@ def _provision_resource(self, context, weighted_host, request_spec,
                         'scheduler.run_instance.scheduled', notifier.INFO,
                         payload)
 
-        updated_instance = driver.instance_update_db(context,
-                instance_uuid, weighted_host.host_state.host)
+        updated_instance = driver.instance_update_db(context, instance_uuid)
 
         self.compute_rpcapi.run_instance(context, instance=updated_instance,
                 host=weighted_host.host_state.host,

nova/tests/compute/test_resource_tracker.py

@@ -22,6 +22,7 @@
 from nova.compute import resource_tracker
 from nova.compute import task_states
 from nova.compute import vm_states
+from nova import context
 from nova import db
 from nova import exception
 from nova.openstack.common import log as logging
@@ -32,14 +33,6 @@
 LOG = logging.getLogger(__name__)
 
 
-class FakeContext(object):
-    def __init__(self, is_admin=False):
-        self.is_admin = is_admin
-
-    def elevated(self):
-        return FakeContext(is_admin=True)
-
-
 class UnsupportedVirtDriver(driver.ComputeDriver):
     """Pretend version of a lame virt driver"""
     def get_available_resource(self):
@@ -81,11 +74,13 @@ def setUp(self):
         self.flags(reserved_host_disk_mb=0,
                    reserved_host_memory_mb=0)
 
-        self.context = FakeContext()
+        self.context = context.RequestContext('fake', 'fake')
 
-        self._instances = []
+        self._instances = {}
         self.stubs.Set(db, 'instance_get_all_by_host',
-                       lambda c, h: self._instances)
+                       lambda c, h: self._instances.values())
+        self.stubs.Set(db, 'instance_update_and_get_original',
+                       self._fake_instance_update_and_get_original)
 
     def _create_compute_node(self, values=None):
         compute = {
@@ -122,8 +117,10 @@ def _create_service(self, host="fakehost", compute=None):
         return service
 
     def _fake_instance(self, *args, **kwargs):
+
+        instance_uuid = str(uuid.uuid1())
         instance = {
-            'uuid': str(uuid.uuid1()),
+            'uuid': instance_uuid,
             'vm_state': vm_states.BUILDING,
             'task_state': None,
             'memory_mb': 2,
@@ -136,9 +133,17 @@ def _fake_instance(self, *args, **kwargs):
         }
         instance.update(kwargs)
 
-        self._instances.append(instance)
+        self._instances[instance_uuid] = instance
         return instance
 
+    def _fake_instance_update_and_get_original(self, context, instance_uuid,
+            values):
+        instance = self._instances[instance_uuid]
+        instance.update(values)
+        # the test doesn't care what the original instance values are, it's
+        # only used in the subsequent notification:
+        return (instance, instance)
+
     def _tracker(self, unsupported=False):
         host = "fakehost"
 
@@ -168,7 +173,8 @@ def testDisabled(self):
 
     def testDisabledClaim(self):
         # basic claim:
-        claim = self.tracker.begin_resource_claim(self.context, 1, 1)
+        instance = self._fake_instance()
+        claim = self.tracker.begin_resource_claim(self.context, instance)
         self.assertEqual(None, claim)
 
     def testDisabledInstanceClaim(self):
@@ -200,7 +206,7 @@ def testDisabledUpdateUsage(self):
 class MissingServiceTestCase(BaseTestCase):
     def setUp(self):
         super(MissingServiceTestCase, self).setUp()
-        self.context = FakeContext(is_admin=True)
+        self.context = context.get_admin_context()
         self.tracker = self._tracker()
 
     def testMissingService(self):

nova/tests/scheduler/test_chance_scheduler.py

@@ -90,8 +90,8 @@ def inc_launch_index(*args):
         self.driver.hosts_up(ctxt_elevated, 'compute').AndReturn(
                 ['host1', 'host2', 'host3', 'host4'])
         random.random().AndReturn(.5)
-        driver.instance_update_db(ctxt, instance1['uuid'],
-                'host3').WithSideEffects(inc_launch_index).AndReturn(instance1)
+        driver.instance_update_db(ctxt, instance1['uuid']).WithSideEffects(
+                inc_launch_index).AndReturn(instance1)
         compute_rpcapi.ComputeAPI.run_instance(ctxt, host='host3',
                 instance=instance1, requested_networks=None,
                 injected_files=None, admin_password=None, is_first_time=None,
@@ -102,8 +102,8 @@ def inc_launch_index(*args):
         self.driver.hosts_up(ctxt_elevated, 'compute').AndReturn(
                 ['host1', 'host2', 'host3', 'host4'])
         random.random().AndReturn(.2)
-        driver.instance_update_db(ctxt, instance2['uuid'],
-                'host1').WithSideEffects(inc_launch_index).AndReturn(instance2)
+        driver.instance_update_db(ctxt, instance2['uuid']).WithSideEffects(
+                inc_launch_index).AndReturn(instance2)
         compute_rpcapi.ComputeAPI.run_instance(ctxt, host='host1',
                 instance=instance2, requested_networks=None,
                 injected_files=None, admin_password=None, is_first_time=None,

nova/tests/scheduler/test_scheduler.py

@@ -712,7 +712,7 @@ def test_cast_to_compute_host_update_db_with_instance_uuid(self):
 
         timeutils.utcnow().AndReturn('fake-now')
         db.instance_update(self.context, 'fake_uuid',
-                {'host': host, 'scheduled_at': 'fake-now'})
+                {'host': None, 'scheduled_at': 'fake-now'})
         rpc.queue_get_for(self.context, 'compute', host).AndReturn(queue)
         rpc.cast(self.context, queue,
                 {'method': method,