When I issue /etc/init.d/zabbix-server stop, start-stop-daemon kills remaining processes while they makes some operations on exit (synchronizing data into DB, flush caches etc). In result openrc marks process as crashed instead correctly stopped. If there HA cluster configuration, ssd kills all processes except "zabbix-server ha manager" process, which prevents correctly start zabbix-server on restarting. There need to introduce --retry option for start-stop-daemon on stop() action for correctly stopping service.