Hyper-V: console logging stops if VM reboots by itself

[Lucian Petrut]

The Hyper-V driver restart VM log workers only in case the instance power changes are triggered by Nova.

As the driver is unaware of VMs that reboot themselves, the VM logging will be stopped in the moment the according seial port named pipe is closed.

The fix consists in emiting and handling instance lifecycle events.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/163080

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/163080
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=e075897a46cf16f10bd70d4a39969e823cf64141
Submitter: Jenkins
Branch: master

commit e075897a46cf16f10bd70d4a39969e823cf64141
Author: Lucian Petrut <email address hidden>
Date: Tue Mar 10 10:51:36 2015 +0200

    Hyper-V: restart serial console workers after instance power change

    At the moment, the only way the Hyper-V handles unexpected power
    state changes is via the manager periodic task. The issue is that
    unexpected instance reboots are not handled at all.

    After an instance is stopped, suspended or paused, the console log
    workers are stopped as well as the serial console pipe is closed.
    Those workers are never restarted unless a power change is triggered
    by the manager.

    This patch proposes an instance state event listener, emiting events
    that can be handled by the manager, similar to the Libvirt driver
    workflow.

    Change-Id: Ia0a875707a7a37224efb333feab1503e83449cc4
    Closes-Bug: #1428505