max failure percentage doesn't work if the node is unreachable from the start

[Alex Schultz]

If you attempt to deploy 3 nodes and 1 is unreachable from the start, the deployment will proceed on the available nodes but will error at the end. We should be failing fast if a node is unavailable and the max_fail_percentage for all roles expects 0 failures.

[Alex Schultz (alex-schultz)]

https://review.opendev.org/c/openstack/tripleo-heat-templates/+/767760

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: stable/train
Review: https://review.opendev.org/c/openstack/tripleo-heat-templates/+/800406

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/tripleo-heat-templates/+/800406
Committed: https://opendev.org/openstack/tripleo-heat-templates/commit/b8d80df42f4e82a7554f9a35a9ab9fde6590c13f
Submitter: "Zuul (22348)"
Branch: stable/train

commit b8d80df42f4e82a7554f9a35a9ab9fde6590c13f
Author: Alex Schultz <email address hidden>
Date: Fri Dec 18 07:55:43 2020 -0700

    Fix unreachable handling

    When we clear the cached facts with unreachable nodes, we attempt to
    gather facts by default. This can cause the node to be skipped for every
    future playbook. This ends up bypassing all our failure percentage
    logic.

    CONFLICT: was lines 15,52. This is changed in Ussuri. To maintain
    consistency with Train, I have kept the Train method for gathering
    facts and just backported the gather_facts: false for the
    "Clear cached facts" task.

    Change-Id: Ie240877496b73a37f553a84af47dfebdbaf899e5
    Related-Bug: 1908573
    (cherry picked from commit 969693e667aba8d893f10f7b03bafde3b3f66287)
    (cherry picked from commit c9ffe726f70e08f8fa564a0d86bcea01f348546c)