Description of problem:
When we update the ceph-ansible configuration and run a stack update, we expect to see all the Ceph OSDs being restarted in order to pick up the new configuration. In RHOSP 16.1 this is what we always observed. However in 16.2 this does not happen, and in order for changes to be picked up we must manually restart the containers.

Version-Release number of selected component (if applicable):
OSP 16.2.3
ceph-ansible-4.0.70.3-1.el8cp.noarch
ceph containers: rhceph/rhceph-4-rhel8 4-69.1638383142 ceph-base-14.2.11-208

How reproducible:
change a osd memory config by adding a template that uses CephAnsibleExtraConfig and osd_memory_target ; then do a deployment with that

parameter_defaults:
  CephAnsibleExtraConfig:
    osd_memory_target: 6123456789


Actual results:
despite the fact that after the deployment changes ar did at /etc/ceph/ceph.conf , new values for "osd memory target" are not applied and needed osd container to restart to having effect


Expected results:
after the deployment new config for osd should be active without their container restart, or some indications in the docs are needed if this is assumed to be an expected result


Additional info: