nova-compute got exception cause of residual instances

[Eric Xie]

Description
===========
Nova-compute got 'KeyError' exception when there is residual instance on the compute node.

Steps to reproduce
==================
* Stop nova-compute service on the compute node
* Then nova reset-state instance on the controller node
* Then nova delete instance on the controller node
* Finally start nova-compute service on the compute node

Expected result
===============
nova-compute service can update its info to DB.

Actual result
=============
nova-compute service got the 'KeyError' exception.

Environment
===========
1. nova version used: stable/mitaka
openstack-nova-scheduler-13.1.0-1.el7.noarch
openstack-nova-compute-13.1.0-1.el7.noarch
openstack-nova-common-13.1.0-1.el7.noarch
openstack-nova-conductor-13.1.0-1.el7.noarch
python-nova-13.1.0-1.el7.noarch
openstack-nova-api-13.1.0-1.el7.noarch
python-novaclient-3.3.1-1.el7.noarch
openstack-nova-console-13.1.0-1.el7.noarch
openstack-nova-novncproxy-13.1.0-1.el7.noarch

2.Libvirt + KVM
libvirt-1.2.17-13.el7_2.5.x86_64
qemu-kvm-ev-2.3.0-31.el7.16.1.x86_64

Logs & Configs
==============
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager Traceback (most recent call last):
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager File "/usr/lib/python2.7/site-packages/nova/compute/manager.py", line 6449, in update_available_resource_for_node
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager rt.update_available_resource(context)
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager File "/usr/lib/python2.7/site-packages/nova/compute/resource_tracker.py", line 508, in update_available_resource
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager resources = self.driver.get_available_resource(self.nodename)
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager File "/usr/lib/python2.7/site-packages/nova/virt/libvirt/driver.py", line 5407, in get_available_resource
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager disk_over_committed = self._get_disk_over_committed_size_total()
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager File "/usr/lib/python2.7/site-packages/nova/virt/libvirt/driver.py", line 7070, in _get_disk_over_committed_size_total
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager local_instances[guest.uuid], bdms[guest.uuid])
2016-10-27 16:40:20.936 25310 ERROR nova.compute.manager KeyError: 'ba0fb33a-bc61-4286-8820-5ee6271ff395'

[Eric Xie (eric-xie)]

I checked the source code and found that:
nova/virt/libvirt/driver.py line 7070
                    block_device_info = driver.get_block_device_info(
                        local_instances[guest.uuid], bdms[guest.uuid])

Should it use local_instances.get(guest.uuid)?