Merge pull request #144 from stackhpc/nova-bug-2091033-antelope

libvirt: Wrap un-proxied listDevices() and listAllDevices()

Author: priteau

nova/tests/unit/virt/libvirt/test_host.py

@@ -2152,6 +2152,48 @@ def test_tpool_list_all_connections(self):
             self.assertIsInstance(domain, tpool.Proxy)
             self.assertIn(domain.UUIDString(), (uuids.vm1, uuids.vm2))
 
+    def _add_fake_host_devices(self):
+        self.conn._obj.pci_info = fakelibvirt.HostPCIDevicesInfo(
+            num_pci=1, num_pfs=2, num_vfs=2, num_mdevcap=3)
+        mdevs = {
+            'mdev_4b20d080_1b54_4048_85b3_a6a62d165c01':
+                fakelibvirt.FakeMdevDevice(
+                    dev_name='mdev_4b20d080_1b54_4048_85b3_a6a62d165c01',
+                    type_id=fakelibvirt.NVIDIA_11_VGPU_TYPE,
+                    parent=fakelibvirt.MDEVCAP_DEV1_PCI_ADDR),
+        }
+        self.conn._obj.mdev_info = fakelibvirt.HostMdevDevicesInfo(
+            devices=mdevs)
+
+    def test_tpool_list_all_devices(self):
+        self._add_fake_host_devices()
+        devs = self.host.list_all_devices(
+            fakelibvirt.VIR_CONNECT_LIST_NODE_DEVICES_CAP_PCI_DEV)
+        self.assertEqual(8, len(devs))
+        for dev in devs:
+            self.assertIsInstance(dev, tpool.Proxy)
+
+    def test_tpool_list_pci_devices(self):
+        self._add_fake_host_devices()
+        devs = self.host.list_pci_devices()
+        self.assertEqual(8, len(devs))
+        for dev in devs:
+            self.assertIsInstance(dev, tpool.Proxy)
+
+    def test_tpool_list_mdev_capable_devices(self):
+        self._add_fake_host_devices()
+        devs = self.host.list_mdev_capable_devices()
+        self.assertEqual(3, len(devs))
+        for dev in devs:
+            self.assertIsInstance(dev, tpool.Proxy)
+
+    def test_tpool_list_mediated_devices(self):
+        self._add_fake_host_devices()
+        devs = self.host.list_mediated_devices()
+        self.assertEqual(1, len(devs))
+        for dev in devs:
+            self.assertIsInstance(dev, tpool.Proxy)
+
 
 class LoadersTestCase(test.NoDBTestCase):
 

nova/virt/libvirt/host.py

@@ -1592,7 +1592,9 @@ def _list_devices(self, cap, flags=0):
         :returns: a list of virNodeDevice instance
         """
         try:
-            return self.get_connection().listDevices(cap, flags)
+            devs = [self._wrap_libvirt_proxy(dev)
+                    for dev in self.get_connection().listDevices(cap, flags)]
+            return devs
         except libvirt.libvirtError as ex:
             error_code = ex.get_error_code()
             if error_code == libvirt.VIR_ERR_NO_SUPPORT:
@@ -1612,7 +1614,10 @@ def list_all_devices(
         :returns: a list of virNodeDevice xml strings.
         """
         try:
-            return self.get_connection().listAllDevices(flags) or []
+            alldevs = [
+                self._wrap_libvirt_proxy(dev)
+                for dev in self.get_connection().listAllDevices(flags)] or []
+            return alldevs
         except libvirt.libvirtError as ex:
             LOG.warning(ex)
             return []

releasenotes/notes/unproxied-libvirt-list-devices-7cd218c1a33535c9.yaml

@@ -0,0 +1,11 @@
+fixes:
+  - |
+    `Bug #2091033`_: Fixed calls to libvirt ``listDevices()`` and
+    ``listAllDevices()`` from potentially blocking all other greenthreads
+    in ``nova-compute``. Under certain circumstances, it was possible for
+    the ``nova-compute`` service to freeze with all other greenthreads
+    blocked and unable to perform any other activities including logging.
+    This issue has been fixed by wrapping the libvirt ``listDevices()``
+    and ``listAllDevices()`` calls with ``eventlet.tpool.Proxy``.
+
+    .. _Bug #2091033: https://bugs.launchpad.net/nova/+bug/2091033