[root@rhos5-rabbitmq1 ~]# systemctl start rabbitmq-server [root@rhos5-rabbitmq1 ~]# ps ax | grep erlang ... take a note .... [root@rhos5-rabbitmq1 ~]# rabbitmqctl status ... note same error as #1104193 [root@rhos5-rabbitmq1 ~]# ps ax | grep erlang notice the new process 3652 ? S 0:00 /usr/lib64/erlang/erts-5.10.4/bin/epmd -daemon [root@rhos5-rabbitmq1 ~]# systemctl stop rabbitmq-server [root@rhos5-rabbitmq1 ~]# ps ax | grep erlang 3652 ? S 0:00 /usr/lib64/erlang/erts-5.10.4/bin/epmd -daemon
It's not easy in systemd to detect if a service is idle. So far the best advice is to stop epmd@<ip-address>.service and the corresponding socket manually. I believe we should close this ticket as CANTFIX.
See the discussion in fedora-devel on that matter: http://thread.gmane.org/gmane.linux.redhat.fedora.devel/201512/focus=201540
You can use: [Service] ExecStop=.... in order to use an external script to cleanup after you. I think there are also other keywords to trigger scripts or commands Before or After Stop.