ceph-ansible/roles/ceph-defaults/templates/restart_mgr_daemon.sh.j2

#!/bin/bash

RETRIES="{{ handler_health_mgr_check_retries }}"
DELAY="{{ handler_health_mgr_check_delay }}"
MGR_NAME="{{ ansible_hostname }}"
SOCKET=/var/run/ceph/{{ cluster }}-mgr.${MGR_NAME}.asok
{% if containerized_deployment %}
DOCKER_EXEC="docker exec ceph-mgr-{{ ansible_hostname }}"
{% endif %}

systemctl reset-failed ceph-mgr@${MGR_NAME}
# First, restart the daemon
systemctl restart ceph-mgr@${MGR_NAME}

COUNT=10
# Wait and ensure the socket exists after restarting the daemds
while [ $RETRIES -ne 0 ]; do
  $DOCKER_EXEC test -S $SOCKET && exit 0
  sleep $DELAY
  let RETRIES=RETRIES-1
done
# If we reach this point, it means the socket is not present.
echo "Socket file ${SOCKET} could not be found, which means ceph manager is not running."
exit 1