# Example: managing cinder controller services with pacemaker # # By setting enabled to false, these services will not be started at boot. By setting # manage_service to false, puppet will not kill these services on every run. This # allows the Pacemaker resource manager to dynamically determine on which node each # service should run. # # The puppet commands below would ideally be applied to at least three nodes. # # Note that cinder-api is associated with the virtual IP address as # it is called from external services. The remaining services connect to the # database and/or message broker independently. # # Example pacemaker resource configuration commands (configured once per cluster): # # sudo pcs resource create cinder_vip ocf:heartbeat:IPaddr2 params ip=192.0.2.3 \ # cidr_netmask=24 op monitor interval=10s # # sudo pcs resource create cinder_api_service lsb:openstack-cinder-api # sudo pcs resource create cinder_scheduler_service lsb:openstack-cinder-scheduler # # sudo pcs constraint colocation add cinder_api_service with cinder_vip class { 'cinder': database_connection => 'mysql://cinder:secret_block_password@openstack-controller.example.com/cinder', } class { 'cinder::api': keystone_password => 'CINDER_PW', keystone_user => 'cinder', enabled => false, manage_service => false, } class { 'cinder::scheduler': enabled => false, manage_service => false, }