Fix inconsistency with ringbuilder/storage steps
authorSteven Hardy <shardy@redhat.com>
Mon, 23 May 2016 16:05:18 +0000 (17:05 +0100)
committerGiulio Fidente <gfidente@redhat.com>
Tue, 31 May 2016 08:59:42 +0000 (10:59 +0200)
commite3cc44579c0e632eb72c3ea9f58f2ab2bc27a251
tree6e327cdb172432207f997e5632c7c9330843ba51
parentd372a3b76faf6860c241b34734d58a3a4c9855fd
Fix inconsistency with ringbuilder/storage steps

Currently when deploying swift on the Controller nodes, we do the
ringbuilder config during step3 and the swift-storage config during
step 4, but this order is reversed on the ObjectStorage nodes.

Also, we include the base swift class inconsistently during step2
on controller nodes, and via the overcloud-object manifest on
ObjectStorage nodes.

So fix this inconsistency as a precursor to conversion to composable
services interfaces for the ObjectStorage role, we rework the post
config so we apply the ObjectStorage config in steps 2, 3 and 4,
which should hopefully get us much closer to the process used
on the controller role, thus be easier to decompose in a compatible
way.

Partially-Implements: blueprint composable-services-within-roles
Change-Id: Ic9d0ed8584a12d681a8f4d4742d39b96c15e531a
puppet/manifests/overcloud_controller.pp
puppet/manifests/overcloud_controller_pacemaker.pp
puppet/manifests/overcloud_object.pp
puppet/manifests/ringbuilder.pp
puppet/swift-storage-post.yaml