Comment 9 for bug 1695904

I gather the "PostPuppet" resources were recently moved to the ControllerPostConfig deployment [0]. The logs show the ControllerPostConfig resource is created [1], but there's no evidence of the ControllerPostPuppetRestart resource being created.

I did some tests using a local deployment, and I checked the deployment plan uploaded to swift on the undercloud. The Jinja2 expanded version of post_puppet_pacemaker.j2.yaml [2] was essentially empty, and did not define any "PostPuppet" resources. I did some more tests, and for reasons I don't understand the check for the 'controller' tag [2] is failing.

I hacked out the check for the 'controller' tag in my deployment, and now the ControllerPostPuppetRestart is being deployed during ControllerPostConfig. I don't understand why the tag check seems broken.

[0] https://review.openstack.org/#/c/459229/
[1] http://logs.openstack.org/76/461076/2/check/gate-tripleo-ci-centos-7-scenario002-multinode-upgrades/cc1f341/console.html#_2017-06-08_19_42_32_498323
[2] https://github.com/openstack/tripleo-heat-templates/blob/stable/ocata/extraconfig/tasks/post_puppet_pacemaker.j2.yaml#L14