From cdf8ba5acb7f65042af9c21fcbe1a126bd857ad0 Mon Sep 17 00:00:00 2001
From: Matt Riedemann <mriedem.os@gmail.com>
Date: Tue, 31 Jul 2018 11:20:47 -0400
Subject: [PATCH] Handle binding_failed vif plug errors on compute restart

Like change Ia584dba66affb86787e3069df19bd17b89cb5c49 which
came before, is port binding fails and we have a "binding_failed"
vif type in the info cache, we'll fail to plug vifs for an
instance on compute restart which will prevent the service
from restarting. Before the os-vif conversion code, this was
handled with VirtualInterfacePlugException but the os-vif conversion
code fails in a different way by raising a generic NovaException
because the os-vif conversion utility doesn't handle a vif_type of
"binding_failed".

To resolve this and make the os-vif flow for binding_failed behave
the same as the legacy path, we implement a translation function
in os_vif_util for binding_failed which will make the plug_vifs
code raise VirtualInterfacePlugException which is what the
_init_instance code in ComputeManager is already handling.

Admittedly this isn't the smartest thing and doesn't attempt
to recover / fix the instance networking info, but it at least
gives a more clear indication of what's wrong and lets the
nova-compute service start up. A note is left in the
_init_instance error handling that we could potentially try
to heal binding_failed vifs in _heal_instance_info_cache, but
that would need to be done in a separate change since it's more
invasive.

Change-Id: Ia963a093a1b26d90b4de2e8fc623031cf175aece
Closes-Bug: #1784579
---
 nova/compute/manager.py                     | 11 +++++++++--
 nova/network/os_vif_util.py                 | 14 ++++++++++++++
 nova/tests/unit/network/test_os_vif_util.py | 18 +++++++++++++++---
 3 files changed, 38 insertions(+), 5 deletions(-)

diff --git a/nova/compute/manager.py b/nova/compute/manager.py
index 0e31263a67a..326b7f9dec1 100644
--- a/nova/compute/manager.py
+++ b/nova/compute/manager.py
@@ -959,8 +959,15 @@ def _init_instance(self, context, instance):
         except NotImplementedError as e:
             LOG.debug(e, instance=instance)
         except exception.VirtualInterfacePlugException:
-            # we don't want an exception to block the init_host
-            LOG.exception("Vifs plug failed", instance=instance)
+            # NOTE(mriedem): If we get here, it could be because the vif_type
+            # in the cache is "binding_failed". The only way to fix that is to
+            # try and bind the ports again, which would be expensive here on
+            # host startup. We could add a check to _heal_instance_info_cache
+            # to handle this, but probably only if the instance task_state is
+            # None.
+            LOG.exception('Virtual interface plugging failed for instance. '
+                          'The port binding:host_id may need to be manually '
+                          'updated.', instance=instance)
             self._set_instance_obj_error_state(context, instance)
             return
 
diff --git a/nova/network/os_vif_util.py b/nova/network/os_vif_util.py
index 1634fb36ff4..e00845737e1 100644
--- a/nova/network/os_vif_util.py
+++ b/nova/network/os_vif_util.py
@@ -481,6 +481,20 @@ def _nova_to_osvif_vif_hostdev_physical(vif):
     raise NotImplementedError()
 
 
+# VIF_TYPE_BINDING_FAILED = 'binding_failed'
+def _nova_to_osvif_vif_binding_failed(vif):
+    """Special handler for the "binding_failed" vif type.
+
+    The "binding_failed" vif type indicates port binding to a host failed
+    and we are trying to plug the vifs again, which will fail because we
+    do not know the actual real vif type, like ovs, bridge, etc. We raise
+    NotImplementedError to indicate to the caller that we cannot handle
+    this type of vif rather than the generic "Unsupported VIF type" error
+    in nova_to_osvif_vif.
+    """
+    raise NotImplementedError()
+
+
 def nova_to_osvif_vif(vif):
     """Convert a Nova VIF model to an os-vif object
 
diff --git a/nova/tests/unit/network/test_os_vif_util.py b/nova/tests/unit/network/test_os_vif_util.py
index 1864f99a48c..203cf83cf9a 100644
--- a/nova/tests/unit/network/test_os_vif_util.py
+++ b/nova/tests/unit/network/test_os_vif_util.py
@@ -14,6 +14,7 @@
 
 from os_vif import objects as osv_objects
 from os_vif.objects import fields as os_vif_fields
+import six
 
 from nova import exception
 from nova.network import model
@@ -1013,9 +1014,20 @@ def test_nova_to_osvif_vif_unknown(self):
                 subnets=[]),
         )
 
-        self.assertRaises(exception.NovaException,
-                          os_vif_util.nova_to_osvif_vif,
-                          vif)
+        ex = self.assertRaises(exception.NovaException,
+                               os_vif_util.nova_to_osvif_vif, vif)
+        self.assertIn('Unsupported VIF type wibble', six.text_type(ex))
+
+    def test_nova_to_osvif_vif_binding_failed(self):
+        vif = model.VIF(
+            id="dc065497-3c8d-4f44-8fb4-e1d33c16a536",
+            type="binding_failed",
+            address="22:52:25:62:e2:aa",
+            network=model.Network(
+                id="b82c1929-051e-481d-8110-4669916c7915",
+                label="Demo Net",
+                subnets=[]),)
+        self.assertIsNone(os_vif_util.nova_to_osvif_vif(vif))
 
     def test_nova_to_osvif_vhostuser_vrouter(self):
         vif = model.VIF(