reload: use OnRestart (#1709)

* reload: use OnRestart

Close the listener on OnRestart for health and metrics so the default
setup function can setup the listener when the plugin is "starting up".

Lightly test with some SIGUSR1-ing. Also checked the reload plugin with
this, seems fine:

.com.:1043
.:1043
2018/04/20 15:01:25 [INFO] CoreDNS-1.1.1
2018/04/20 15:01:25 [INFO] linux/amd64, go1.10,
CoreDNS-1.1.1
linux/amd64, go1.10,
2018/04/20 15:01:25 [INFO] Running configuration MD5 = aa8b3f03946fb60546ca1f725d482714
2018/04/20 15:02:01 [INFO] Reloading
2018/04/20 15:02:01 [INFO] Running configuration MD5 = b34a96d99e01db4015a892212560155f
2018/04/20 15:02:01 [INFO] Reloading complete
^C2018/04/20 15:02:06 [INFO] SIGINT: Shutting down

With this corefile:
.com {
  proxy . 127.0.0.1:53
  prometheus :9054
  whoami
  reload
}

. {
  proxy . 127.0.0.1:53
  prometheus :9054
  whoami
  reload
}

The prometheus port was 9053, changed that to 54 so reload would pick it
up.

From a cursory look it seems this also fixes:
Fixes #1604 #1618 #1686 #1492

* At least make it test

* Use onfinalshutdown

* reload: add reload test

This test #1604 adn right now fails.

* Address review comments

* Add bug section explaining things a bit

* compile tests

* Fix tests

* fixes

* slightly less crazy

* try to make prometheus setup less confusing

* Use ephermal port for test

* Don't use the listener

* These are shared between goroutines, just use the boolean in the main
  structure.
* Fix text in the reload README,
* Set addr to TODO once stopping it
* Morph fturb's comment into test, to test reload and scrape health and
  metric endpoint

plugin/health/README.md

@@ -17,9 +17,9 @@ health [ADDRESS]
 ~~~
 
 Optionally takes an address; the default is `:8080`. The health path is fixed to `/health`. The
-health endpoint returns a 200 response code and the word "OK" when CoreDNS is healthy. It returns
-a 503. *health* periodically (1s) polls plugin that exports health information. If any of the
-plugin signals that it is unhealthy, the server will go unhealthy too. Each plugin that supports
+health endpoint returns a 200 response code and the word "OK" when this server is healthy. It returns
+a 503. *health* periodically (1s) polls plugins that exports health information. If any of the
+plugins signals that it is unhealthy, the server will go unhealthy too. Each plugin that supports
 health checks has a section "Health" in their README.
 
 More options can be set with this extended syntax:
@@ -33,7 +33,7 @@ health [ADDRESS] {
 * Where `lameduck` will make the process unhealthy then *wait* for **DURATION** before the process
   shuts down.
 
-If you have multiple Server Block and need to export health for each of the plugins, you must run
+If you have multiple Server Blocks and need to export health for each of the plugins, you must run
 health endpoints on different ports:
 
 ~~~ corefile

plugin/health/health.go

@@ -16,8 +16,9 @@ type health struct {
 	Addr     string
 	lameduck time.Duration
 
-	ln  net.Listener
-	mux *http.ServeMux
+	ln      net.Listener
+	nlSetup bool
+	mux     *http.ServeMux
 
 	// A slice of Healthers that the health plugin will poll every second for their health status.
 	h []Healther
@@ -45,6 +46,7 @@ func (h *health) OnStartup() error {
 
 	h.ln = ln
 	h.mux = http.NewServeMux()
+	h.nlSetup = true
 
 	h.mux.HandleFunc(path, func(w http.ResponseWriter, r *http.Request) {
 		if h.Ok() {
@@ -61,7 +63,13 @@ func (h *health) OnStartup() error {
 	return nil
 }
 
-func (h *health) OnShutdown() error {
+func (h *health) OnRestart() error { return h.OnFinalShutdown() }
+
+func (h *health) OnFinalShutdown() error {
+	if !h.nlSetup {
+		return nil
+	}
+
 	// Stop polling plugins
 	h.pollstop <- true
 	// NACK health
@@ -72,11 +80,10 @@ func (h *health) OnShutdown() error {
 		time.Sleep(h.lameduck)
 	}
 
-	if h.ln != nil {
-		return h.ln.Close()
-	}
+	h.ln.Close()
 
 	h.stop <- true
+	h.nlSetup = false
 	return nil
 }
 

plugin/health/health_test.go

@@ -17,7 +17,7 @@ func TestHealth(t *testing.T) {
 	if err := h.OnStartup(); err != nil {
 		t.Fatalf("Unable to startup the health server: %v", err)
 	}
-	defer h.OnShutdown()
+	defer h.OnFinalShutdown()
 
 	go func() {
 		<-h.pollstop
@@ -73,5 +73,5 @@ func TestHealthLameduck(t *testing.T) {
 		return
 	}()
 
-	h.OnShutdown()
+	h.OnFinalShutdown()
 }

plugin/health/setup.go

@@ -60,7 +60,8 @@ func setup(c *caddy.Controller) error {
 	})
 
 	c.OnStartup(h.OnStartup)
-	c.OnShutdown(h.OnShutdown)
+	c.OnRestart(h.OnRestart)
+	c.OnFinalShutdown(h.OnFinalShutdown)
 
 	// Don't do AddPlugin, as health is not *really* a plugin just a separate webserver running.
 	return nil