The allocation table has residual records when instance is evacuated and the source node is removed

[Sun Mengyun]

Description
===========
When compute service is down, we choose to evacuate the instance of corresponding node. If success, there will be two records in allocations table. Normally, we try to restore the nova-compute service and one of the records will be deleted when nova-compute service is restarted. However, if the node fails unrecoverably and becomes unavailable, the compute service will never restart. Eventually, the records in the table will remain.

Further more， if we delete the compute service in failed node, the corresponding records in resource_provider table will not be deleted too, because the remained record in allocation table references this record.

Ultimately, we will have at least two residual records.

Steps to reproduce
==================

1. Let compute service down
2. evacuate the instance on the node with commond：
    nova evacuate instance_uuid
3.delete the service
    nova service-delete service_id

Expected result
===============
The corrsponding records in allocations and resource_provider are deleted

Actual result
=============
You will find two records in allocation table and one them is invalid, ditto one in resource_provider table

Environment
===========
[root@nail1 ~]# rpm -qa | grep nova
openstack-nova-api-18.0.2-1.el7.noarch
openstack-nova-common-18.0.2-1.el7.noarch
python2-novaclient-11.0.0-1.el7.noarch
openstack-nova-placement-api-18.0.2-1.el7.noarch
openstack-nova-scheduler-18.0.2-1.el7.noarch
openstack-nova-conductor-18.0.2-1.el7.noarch
openstack-nova-novncproxy-18.0.2-1.el7.noarch
python-nova-18.0.2-1.el7.noarch
openstack-nova-compute-18.0.2-1.el7.noarch
openstack-nova-console-18.0.2-1.el7.noarch

[Matt Riedemann (mriedem)]

This sounds like a duplicate of bug 1724172.

[Sun Mengyun (kmehxhcr)]

Yes, the same question.