Swift overcloud rings inconsistency in containerized environment

[Christian Schwede]

The Swift rings on the overcloud nodes are inconsistent when using a containerized environment in Tripleo. This happens when nodes are replaced or added to an existing deployment later.

Rings are neither fetched from the undercloud nor stored on the undercloud after rebalancing, which is due to the fact that these actions are tied to step 2 & 5 in puppet-tripleo, but the config is executed within step 6 only.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/468790

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/468790
Committed: https://git.openstack.org/cgit/openstack/puppet-tripleo/commit/?id=410e05ba63cad12d99f33717d116213a38413740
Submitter: Jenkins
Branch: master

commit 410e05ba63cad12d99f33717d116213a38413740
Author: Christian Schwede <email address hidden>
Date: Mon May 29 09:56:55 2017 +0200

    Fix Swift ring management in container deployments

    The ring up- and downloading was never executed if run within a
    containerized environment. This is due to the fact that this manifest
    gets executed within step 6(5) only. There is also an ordering issue,
    which actually tries to create the tarballs before rebalancing.

    This patch fixes the step conditions and also chains the tarball
    creation to the rebalance.

    The check to query rings on all nodes can now be disabled. This is
    required on containerized environments: the local ring will be modified
    and rebalanced, but rings on the existing servers are not yet modified.
    Therefore a recon-check will fail, and needs to be disabled.

    Closes-Bug: 1694211
    Change-Id: I51c5795b9893d797bd73e059910f17a98f04cdbe

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/puppet-tripleo 7.2.0 release.