openstack/nova
Code Issues Proposed changes

Return HTTP 200 on list for invalid status

Browse Source 
The server listing API raises a 500 error if you pass an invalid status
filter for admin user. In the case of a non-admin user, it simply
returns an empty list. In the case of an admin user, it fetches extended
server attributes, so a condition was added to get extended server
attributes only when servers list is not empty.

This change simply removes the cause of the 500 exception. A subsequent
patch with a microversion bump will modify the behavior so that a 400
Bad Request will be raised for an invalid status, for both admin and
non-admin alike.

Co-Authored-By: Dinesh Bhor <dinesh.bhor@nttdata.com>

Closes-Bug: #1579706

Change-Id: I10bde78f0a9ac59b8646d58f62fa5056f989f54f
This commit is contained in:
EdLeafe 2016-06-29 18:51:34 +00:00
parent 222a88a54c
commit ee4d69e28d
3 changed files with 33 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
nova/api/openstack/compute/extended_server_attributes.py
View File

@ -91,18 +91,21 @@ class ExtendedServerAttributesController(wsgi.Controller):
authorize_host_status = True
if authorize_extend or authorize_host_status:
servers = list(resp_obj.obj['servers'])
instances = req.get_db_instances()
# Instances is guaranteed to be in the cache due to
# the core API adding it in its 'detail' method.
if authorize_host_status:
host_statuses = self.compute_api.get_instances_host_statuses(
instances.values())
for server in servers:
if authorize_extend:
instance = instances[server['id']]
self._extend_server(context, server, instance, req)
# NOTE(dinesh-bhor): Skipping fetching of instances from cache as
# servers list can be empty if invalid status is provided to the
# core API 'detail' method.
if servers:
instances = req.get_db_instances()
if authorize_host_status:
server['host_status'] = host_statuses[server['id']]
host_statuses = (
self.compute_api.get_instances_host_statuses(
instances.values()))
for server in servers:
if authorize_extend:
instance = instances[server['id']]
self._extend_server(context, server, instance, req)
if authorize_host_status:
server['host_status'] = host_statuses[server['id']]
class ExtendedServerAttributes(extensions.V21APIExtensionBase):

12
nova/tests/unit/api/openstack/compute/test_extended_server_attributes.py
View File

@ -13,6 +13,7 @@
# License for the specific language governing permissions and limitations
# under the License.
import mock
from oslo_config import cfg
from oslo_serialization import jsonutils
@ -129,6 +130,17 @@ class ExtendedServerAttributesTestV21(test.TestCase):
node='node-%s' % (i + 1),
instance_name=NAME_FMT % (i + 1))
@mock.patch.object(compute.api.API, 'get_all')
def test_detail_empty_instance_list_invalid_status(self,
mock_get_all_method):
mock_get_all_method.return_value = objects.InstanceList(objects=[])
url = "%s%s" % (self.fake_url, '/servers/detail?status=invalid_status')
res = self._make_request(url)
# check status code 200 with empty instance list
self.assertEqual(200, res.status_int)
self.assertEqual(0, len(self._get_servers(res.body)))
def test_no_instance_passthrough_404(self):
def fake_compute_get(*args, **kwargs):

7
releasenotes/notes/list-server-bad-status-fix-7db504b38c8d732f.yaml Normal file
View File

@ -0,0 +1,7 @@
---
fixes:
- |
Fixed bug #1579706: "Listing nova instances with invalid status raises 500
InternalServerError for admin user". Now passing an invalid status as a
filter will return an empty list. A subsequent patch will then correct this
to raise a 400 Bad Request when an invalid status is received.