run systemctl as the init process

README

@@ -0,0 +1,48 @@
+If this 'systemctl' is run without arguments
+then it will default to "list-units".
+
+## systemctl as init process
+
+However... if systemctl is PID 1 then it will
+run "1" which does both "default" and "wait".
+So it will start all services in the default
+target of "multi-user.target", it will then
+wait for any child to be reaped. And when a
+SIGTERM (aka "docker stop") is received then
+it will stop all services in the default 
+target of "multi-user.target".
+
+As recommendations, you may try to run
+
+    docker run --name running image /usr/bin/systemctl
+
+and later
+
+    docker stop --time 100 running
+
+Note that run does have a "--stop-timeout 100" in
+later versions corresponding to docker-compse
+stop_grace_period setting. In most cases the default
+of 10 seconds is to short to bring down all services
+correctly. Make it atleast 100 if you know that this
+systemctl replacement is run as PID 1 .. i.e. as a
+"init" process replacement.
+
+You may force the docker-systemctl-replacement to
+run as the init replacement simply by saying
+
+    /usr/bin/systemctl 1
+
+which is identical to
+
+    /usr/bin/systemctl default
+    /usr/bin/systemctl wait
+
+The handling of SIGTERM in "wait" is to call "halt"
+which may also be issued manually to bring down
+all services:
+
+    /usr/bin/systemctl halt
+
+The systemctl commands "default" and "halt" are matching
+those from the original, while "wait" is an innovation.