Description of problem:

The OpenStack has been deployed by the OpenStack deployer while the Ceph cluster has been deployed manually.

In this situation we are hitting a bug in the deployer which is not properly setting up the virsh secret needed for the CephX authentication to work. This behavior is due to a condition in Puppet manifest preventing the deployer to properly set it if the Ceph cluster is already existing and not deployed by the deployer.

/etc/puppet/environments/production/modules/quickstack/manifests/compute_common.pp

    131     exec { 'define-virsh-rbd-secret':
    132       command => '/usr/bin/virsh secret-define --file /etc/nova/secret.xml',
    133       onlyif => "/usr/bin/ceph --connect-timeout 10 auth get-key client.${libvirt_images_rbd_pool} >/dev/null 2>&1",
    134       creates => '/etc/nova/virsh.secret',
    135     }
    136     ->
    137     exec { 'set-virsh-rbd-secret-key':
    138       command => "/usr/bin/virsh secret-set-value --secret ${rbd_secret_uuid} --base64 \$(/usr/bin/ceph auth get-key client.${libvirt_images_rbd_pool})",
    139       onlyif 

# /usr/bin/ceph --connect-timeout 10 auth get-key client.volumes
2015-02-25 16:13:20.144056 7fbfc77ee700 -1 monclient(hunting): ERROR: missing keyring, cannot use cephx for authentication







Version-Release number of selected component (if applicable):
rhel-osp-installer-0.5.5-2.el7ost.noarch
rhel-osp-installer-client-0.5.5-2.el7ost.noarch
ruby193-rubygem-staypuft-0.5.15-1.el7ost.noarch
apisto-puppet-3.4.3-1.1.x86_64  
openstack-puppet-modules-2014.2.8-2.el7ost.noarch
puppet-3.6.2-2.el7.noarch
puppet-server-3.6.2-2.el7.noarch

How reproducible:
Deploy openstack with an already deployed ceph cluster


Steps to Reproduce:
1. Deploy ceph manually
2. Deploy openstack to use ceph
3.

Actual results:
# /usr/bin/ceph --connect-timeout 10 auth get-key client.volumes
2015-02-25 16:13:20.144056 7fbfc77ee700 -1 monclient(hunting): ERROR: missing keyring, cannot use cephx for authentication


Expected results:
Openstack installed using the manually deployed ceph cluster

Additional info: