A health middleware

Start http handler on port 8080 and return OK. Also add some

documentation fixes for the prometheus middleware.

middleware/prometheus/README.md

@@ -9,9 +9,10 @@ The following metrics are exported:
 * coredns_dns_response_size_bytes
 * coredns_dns_response_rcode_count_total
 
-Each counter has a label `zone` which is the zonename used for the request/response.
-The `request_count` metrics has an extra label `qtype` which holds the qtype. And
-`rcode_count` has an extra label which has the rcode.
+Each counter has a label `zone` which is the zonename used for the request/response,
+and a label `qtype` which old the query type.
+The `response_rcode_count_total` has an extra label `rcode` which holds the rcode
+of the response.
 
 Restarting CoreDNS will stop the monitoring. This is a bug. Also [this upstream
 Caddy bug](https://github.com/mholt/caddy/issues/675).
@@ -25,6 +26,6 @@ prometheus
 For each zone that you want to see metrics for.
 
 It optionally takes an address where the metrics are exported, the default
-is `localhost:9154`. The metrics path is fixed to `/metrics`.
+is `localhost:9135`. The metrics path is fixed to `/metrics`.
 
 ## Examples

middleware/prometheus/metrics.go

@@ -1,7 +1,7 @@
 package metrics
 
 import (
-	"fmt"
+	"log"
 	"net/http"
 	"sync"
 
@@ -39,8 +39,9 @@ func (m *Metrics) Start() error {
 
 		http.Handle(path, prometheus.Handler())
 		go func() {
-			// TODO(miek): Logging here?
-			fmt.Errorf("%s", http.ListenAndServe(m.Addr, nil))
+			if err := http.ListenAndServe(m.Addr, nil); err != nil {
+				log.Printf("[ERROR] Failed to start prometheus handler: %s", err)
+			}
 		}()
 	})
 	return nil