No bootable device when evacuate a instance on shared_storage_storage ceph

[sampsonhuo]

Nova Verion：nova-kilo-2015.1.1
Ceph Verion：ceph version 10.2.2 (45107e21c568dd033c2f0a3107dec8f0b0e58374)

When i test nova evacuate function i found after the instance evacuated it cant not boot normally.
By the vnc console i see "No bootable device" info.

Through some tests i found the reason： when u used the shared storage the rebuild task flow will not get the image meta again. So if u set meta for the image, the problem occur.

The code：
 nova/compute/manager.py

 @object_compat
 @messaging.expected_exceptions(exception.PreserveEphemeralNotSupported)
 @wrap_exception()
 @reverts_task_state
 @wrap_instance_event
 @wrap_instance_fault
 def rebuild_instance(self, context, instance, orig_image_ref, image_ref,
   injected_files, new_pass,
      orig_sys_metadata,
   bdms, recreate, on_shared_storage,
   preserve_ephemeral=False):
    ......

if on_shared_storage != self.driver.instance_on_disk(instance):
  raise exception.InvalidSharedStorage(_("Invalid state of instance files on shared"
    " storage"))

if on_shared_storage:
  LOG.info(_LI('disk on shared storage, recreating using'
         ' existing disk'))
else:
  image_ref = orig_image_ref = instance.image_ref
  LOG.info(_LI("disk not on shared storage, rebuilding from:"
         " '%s'"), str(image_ref))

# NOTE(mriedem): On a recreate (evacuate), we need to update
# the instance's host and node properties to reflect it's
# destination node for the recreate.
node_name = None
try:
  compute_node = self._get_compute_info(context, self.host)
     node_name = compute_node.hypervisor_hostname
except exception.ComputeHostNotFound:
  LOG.exception(_LE('Failed to get compute_info for %s'),self.host)
finally:
  instance.host = self.host
  instance.node = node_name
  instance.save()

if image_ref:
  image_meta = self.image_api.get(context, image_ref)
else:
  image_meta = {}
......

Bellow is my image info
+------------------------------+--------------------------------------+
| Property | Value |
+------------------------------+--------------------------------------+
| OS-EXT-IMG-SIZE:size | 53687091200 |
| created | 2016-09-20T08:15:21Z |
| id | 8b218b4d-74ff-44af-bc4c-c37fb1106b03 |
| metadata hw_disk_bus | scsi |
| metadata hw_qemu_guest_agent | yes |
| metadata hw_scsi_model | virtio-scsi |
| minDisk | 0 |
| minRam | 0 |
| name | zptest-20160920 |
| progress | 100 |
| status | ACTIVE |
| updated | 2016-10-20T07:38:54Z |
+------------------------------+--------------------------------------+

[sampsonhuo (13261149612-5)]

I found the same bug:
https://bugs.launchpad.net/nova/+bug/1562681

but in Kilo i update like this:

nova/compute/manager.py
the rebuild_instance function update from:

if on_shared_storage:
  LOG.info(_LI('disk on shared storage, recreating using'
         ' existing disk'))
else:
  image_ref = orig_image_ref = instance.image_ref
  LOG.info(_LI("disk not on shared storage, rebuilding from:"
         " '%s'"), str(image_ref))

to

  image_ref = orig_image_ref = instance.image_ref
  LOG.info(_LI("rebuilding from:'%s'"), str(image_ref))

[Bing Li (libing863)]

This bug seems duplicated with https://bugs.launchpad.net/nova/+bug/1562681, and has been fixe in Liberty and newer versions.