Suppose that your original answer-file been used for Controller/Network +(N)*Compute Node deployment has been updated as follows to separate Storage Node from Controller. Controller has IP 192.169.142.127, Compute has IP 192.169.142.137. Following lines in classic ML2&OVS&VXLAN deployment answer-file are supposed to be updated.
EXCLUDE_SERVERS=192.169.142.127,192.169.142.137
. . . . . . .
CONFIG_UNSUPPORTED=y
. . . . . . .
CONFIG_STORAGE_HOST=192.169.142.117
. . . . . . .
CONFIG_NEUTRON_OVS_TUNNEL_IF=eth1
CONFIG_NEUTRON_OVS_TUNNEL_SUBNETS=12.0.0.0/24
As far as updates above are correct new node 192.169.142.117 will be added to landscape, however endpoints for all storage services in keystone database wouldn’t be updated. So before stopping services on Controller keystone database is supposed to be hacked. Cinder may be tuned manipulating bindings cinder-types to required volume-groupes via cinder.conf on 192.169.142.117.
Complete text may be seen at http://bderzhavets.blogspot.com/2016/04/switching-to-newly-added-storage-node.html
