Keepalived not considered to declare unhealthy an amphorae on reboot

[Leandro Reox]

Doing ACTIVE STANDBY reboot tests, due to a race condition keepalived failed to start but the amphorae wasnt considered as unhealthy so it can be replaced.

This can cause that if we:

1 - Reboot the master node
2 - Keepalived fail to start on master node
3 - Shutdown backup node
4 - Traffic dropped

Keepalived should be considered on a reboot as a health indicator, cause the rebooted node will not claim the VIP again if the backup fails in the middle.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/470081

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/470081
Committed: https://git.openstack.org/cgit/openstack/octavia/commit/?id=c7a2babf545a30ac29f5dda91e19eaca933c106a
Submitter: Jenkins
Branch: master

commit c7a2babf545a30ac29f5dda91e19eaca933c106a
Author: Michael Johnson <email address hidden>
Date: Thu Jun 1 18:55:54 2017 -0700

    Check keepalived health in the amphora

    The health manager did not detect a keepalived failure inside the amphora.
    This patch will not send a health heartbeat if keepalived is configured but
    not running.

    This patch also allows the health checks to continue after an initial failure.

    Change-Id: Id21310bd5ded3747218d3872ab3c966e5ddf5356
    Closes-Bug: #1695090

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/octavia 1.0.0.0b2 development milestone.