fixed ip address assigned twice to the vm

[Ankit Agrawal]

If the vm fails to spawn on the nova compute node where it was suppose to run, then it reschedules to run the vm on the another compute node.
Network of VM is deleted before trying to re-schedule it on another host. But system_metadata of instance is not updated ('network_allocated' is not
set to False). We should also update system_metadata after network is deleted from VM.

While allocating network to instance we set system_metadata 'network_allocated' as True, but while deallocating network we do not set 'network_allocated'
to False.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/75346

[Andrew Laski (alaski)]

Do you have logs or more information on a case where you've seen this happen? I'm trying to understand this but the report seems a bit inconsistent. The title says that the fixed_ip will be assigned twice, but the description says that network_allocated will be true when it shouldn't be which would mean no networking is assigned right?

Also, now that Iefab71047996b7cc08107794d5bc628c11680a70 has merged the network is not deallocated on a reschedule except when using the baremetal driver.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Ankit Agrawal (<email address hidden>) on branch: master
Review: https://review.openstack.org/75346