Combine BootstrapNodeDeployment with AllNodesDeployment
authorSteven Hardy <shardy@redhat.com>
Mon, 4 Jul 2016 15:28:41 +0000 (16:28 +0100)
committerSteven Hardy <shardy@redhat.com>
Tue, 5 Jul 2016 09:58:31 +0000 (10:58 +0100)
commitb345dbea16ad3edd600c62848d8ee116f4df16ee
tree26e0820820da66b88b5380c408f3ed7ab38f1f9c
parent519b94ff1e2b4d3de563ca28c14cb48a4fbcf34c
Combine BootstrapNodeDeployment with AllNodesDeployment

Currently we have a special controller-only deployment which writes
the name/ip of the "bootstrap node", e.g the cluster master, which
defaults to the first node in the Controller ResourceGroup.

Now we're moving to fully composable services/roles, it's possible
folks will want to deploy services that expect to detect the bootstrap
node (e.g so only one node does a DB sync) for non-controller roles.

So, take this opportunity to combine the bootstrap node deployment with
the "all nodes" data, such that we deploy the same data for all roles.

Because the boostrap node data is per role cluster, rather than truly
global, we pass it via input_values into each per-role Deployment.

At some future point we might consider renaming this, e.g to
something which describes per-cluster config vs "all nodes",
but as a first step let's just rationalize the resources.

Change-Id: I4011526a13c51b3d0f95c17fe8ed38115b4fdce4
overcloud-resource-registry-puppet.yaml
overcloud.yaml
puppet/all-nodes-config.yaml
puppet/bootstrap-config.yaml [deleted file]
puppet/ceph-storage.yaml
puppet/cinder-storage.yaml
puppet/swift-storage.yaml