openstack
/
nova
Code Issues Proposed changes

Add populate_retry to schedule_and_build_instances 

When boot an instance and failed on the compute node, nova will
not retry to boot on other host.

Since https://review.openstack.org/#/c/319379/ change the create
instance workflow and called schedule_and_build_instances which
not populate the retry into filter properties. So nova will not
retry when boot on compute fail. This patch populate retry to
instance properties when call schedule_and_build_instances.

Conflicts:
	nova/tests/functional/regressions/test_bug_1671648.py

NOTE(mriedem): The conflict is due to putting the functional
test fix before this bug fix in the backport series.

Change-Id: Ifdaddcd265a7fe8282499e27043936f8212610ad
Closes-Bug: #1671648
(cherry picked from commit cb4ce72f5f)
This commit is contained in:
ShunliZhou 2017-03-10 14:05:57 +08:00 committed by Matt Riedemann
parent 6304edf608
commit 4e3be434bd
3 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
nova/conductor/manager.py
View File

@ -875,9 +875,10 @@ class ComputeTaskManager(base.Base):
for (build_request, request_spec, host) in six.moves.zip(
build_requests, request_specs, hosts):
filter_props = request_spec.to_legacy_filter_properties_dict()
instance = build_request.get_new_instance(context)
scheduler_utils.populate_retry(filter_props, instance.uuid)
scheduler_utils.populate_filter_properties(filter_props,
host)
instance = build_request.get_new_instance(context)
# Convert host from the scheduler into a cell record
if host['host'] not in host_mapping_cache:

19
nova/tests/functional/regressions/test_bug_1671648.py
View File

@ -136,22 +136,19 @@ class TestRetryBetweenComputeNodeBuilds(test.TestCase):
scheduler picks the first host which we mock out to fail the claim.
This should then trigger a retry to the second host.
"""
# Create the server which we expect to go to ERROR state because
# of the regression bug. Once the bug is fixed, we should assert
# that the server goes to ACTIVE status and is on the second host
# after the retry operation.
# Now that the bug is fixed, we should assert that the server goes to
# ACTIVE status and is on the second host after the retry operation.
server = dict(
name='retry-test',
imageRef=self.image_id,
flavorRef=self.flavor_id)
server = self.admin_api.post_server({'server': server})
self.addCleanup(self.admin_api.delete_server, server['id'])
server = self._wait_for_instance_status(server['id'], 'ERROR')
server = self._wait_for_instance_status(server['id'], 'ACTIVE')
# Assert that there is no host for the failed server. This should
# assert that the host is not the failed host once the bug is fixed.
self.assertIsNone(server['OS-EXT-SRV-ATTR:host'])
# Assert that the host is not the failed host.
self.assertNotEqual(self.failed_host,
server['OS-EXT-SRV-ATTR:host'])
# Assert that we did not retry. Once the bug is fixed, this should
# be equal to 2.
self.assertEqual(1, self.attempts)
# Assert that we retried.
self.assertEqual(2, self.attempts)

1
nova/tests/unit/conductor/test_conductor.py
View File

@ -1442,6 +1442,7 @@ class ConductorTaskTestCase(_BaseTaskTestCase, test_compute.BaseTestCase):
def _build_and_run_instance(ctxt, *args, **kwargs):
details['instance'] = kwargs['instance']
self.assertTrue(kwargs['instance'].id)
self.assertTrue(kwargs['filter_properties'].get('retry'))
self.assertEqual(1, len(kwargs['block_device_mapping']))
# FIXME(danms): How to validate the db connection here?