[octane] live migration doesn't work when upgrade is done

[Pavel Chechetin]

Detailed bug description:
 When an upgrade is done () live migration between two compute nodes doesn't work due to inconsistent hosts file on some compute nodes.

Because octane upgrades a compute node in one-by-one fashion all compute nodes (except the last one) have inconsistent /etc/hosts file.

The root cause of such a inconsistency is octane, thought API calls, emulate "fuel node --deploy" command, but it seems that it doesn't update /etc/hosts on other compute nodes.

Steps to reproduce:
 Upgrade env with at least 2 compute nodes.
 Launch a VM
 Live migrate this VM

Expected results:
 VM should disappear from the source compute node and appear on destination.

Actual result:
 VM gets stuck for sometime in migration state and doesn't appear on destination node.

Reproducibility:
 Always

Workaround:
  Launch task hosts on all compute nodes.

Impact:
  Upgrades

Description of the environment:
 Operation system: Ubuntu
 Versions of components: Fuel 8.0, Fuel 9.0 (may be)
 Network model: Tunnel

[Pavel Chechetin (paulche)]

The description wasn't accurate 'cause this issue only occurs when an env is under upgrade procedure. I'll update the description.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/327525

[OpenStack Infra (hudson-openstack)]

Change abandoned by Ilya Kharin (<email address hidden>) on branch: master
Review: https://review.openstack.org/327525
Reason: This thing will be made by a little bit different way.

[Anton Matveev (amatveev)]

octane specific, moving to won't fix due to shifted priorities