plugin/metrics: add 'server' label (#1682)

* plugin/metrics: add 'server' label

This uses the new WithServer(ctx) to get the current server from the
context.

First in a larger refactor to make all plugins do this.

* compile

* compile

* lala test

* compile and test

* typos

* Dont duplicate the code

core/dnsserver/server.go

@@ -197,7 +197,7 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 	// The default dns.Mux checks the question section size, but we have our
 	// own mux here. Check if we have a question section. If not drop them here.
 	if r == nil || len(r.Question) == 0 {
-		DefaultErrorFunc(w, r, dns.RcodeServerFailure)
+		DefaultErrorFunc(ctx, w, r, dns.RcodeServerFailure)
 		return
 	}
 
@@ -206,13 +206,13 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 			// In case the user doesn't enable error plugin, we still
 			// need to make sure that we stay alive up here
 			if rec := recover(); rec != nil {
-				DefaultErrorFunc(w, r, dns.RcodeServerFailure)
+				DefaultErrorFunc(ctx, w, r, dns.RcodeServerFailure)
 			}
 		}()
 	}
 
 	if !s.classChaos && r.Question[0].Qclass != dns.ClassINET {
-		DefaultErrorFunc(w, r, dns.RcodeRefused)
+		DefaultErrorFunc(ctx, w, r, dns.RcodeRefused)
 		return
 	}
 
@@ -223,7 +223,7 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 
 	ctx, err := incrementDepthAndCheck(ctx)
 	if err != nil {
-		DefaultErrorFunc(w, r, dns.RcodeServerFailure)
+		DefaultErrorFunc(ctx, w, r, dns.RcodeServerFailure)
 		return
 	}
 
@@ -254,7 +254,7 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 				if h.FilterFunc == nil {
 					rcode, _ := h.pluginChain.ServeDNS(ctx, w, r)
 					if !plugin.ClientWrite(rcode) {
-						DefaultErrorFunc(w, r, rcode)
+						DefaultErrorFunc(ctx, w, r, rcode)
 					}
 					return
 				}
@@ -263,7 +263,7 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 				if h.FilterFunc(q) {
 					rcode, _ := h.pluginChain.ServeDNS(ctx, w, r)
 					if !plugin.ClientWrite(rcode) {
-						DefaultErrorFunc(w, r, rcode)
+						DefaultErrorFunc(ctx, w, r, rcode)
 					}
 					return
 				}
@@ -285,7 +285,7 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 		// DS request, and we found a zone, use the handler for the query.
 		rcode, _ := dshandler.pluginChain.ServeDNS(ctx, w, r)
 		if !plugin.ClientWrite(rcode) {
-			DefaultErrorFunc(w, r, rcode)
+			DefaultErrorFunc(ctx, w, r, rcode)
 		}
 		return
 	}
@@ -298,13 +298,13 @@ func (s *Server) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 
 		rcode, _ := h.pluginChain.ServeDNS(ctx, w, r)
 		if !plugin.ClientWrite(rcode) {
-			DefaultErrorFunc(w, r, rcode)
+			DefaultErrorFunc(ctx, w, r, rcode)
 		}
 		return
 	}
 
 	// Still here? Error out with REFUSED.
-	DefaultErrorFunc(w, r, dns.RcodeRefused)
+	DefaultErrorFunc(ctx, w, r, dns.RcodeRefused)
 }
 
 // OnStartupComplete lists the sites served by this server
@@ -331,7 +331,7 @@ func (s *Server) Tracer() ot.Tracer {
 }
 
 // DefaultErrorFunc responds to an DNS request with an error.
-func DefaultErrorFunc(w dns.ResponseWriter, r *dns.Msg, rc int) {
+func DefaultErrorFunc(ctx context.Context, w dns.ResponseWriter, r *dns.Msg, rc int) {
 	state := request.Request{W: w, Req: r}
 
 	answer := new(dns.Msg)
@@ -339,7 +339,7 @@ func DefaultErrorFunc(w dns.ResponseWriter, r *dns.Msg, rc int) {
 
 	state.SizeAndDo(answer)
 
-	vars.Report(state, vars.Dropped, rcode.ToString(rc), answer.Len(), time.Now())
+	vars.Report(ctx, state, vars.Dropped, rcode.ToString(rc), answer.Len(), time.Now())
 
 	w.WriteMsg(answer)
 }

plugin/log/log.go

@@ -21,7 +21,7 @@ import (
 type Logger struct {
 	Next      plugin.Handler
 	Rules     []Rule
-	ErrorFunc func(dns.ResponseWriter, *dns.Msg, int) // failover error handler
+	ErrorFunc func(context.Context, dns.ResponseWriter, *dns.Msg, int) // failover error handler
 }
 
 // ServeDNS implements the plugin.Handler interface.
@@ -39,13 +39,13 @@ func (l Logger) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg)
 			// There was an error up the chain, but no response has been written yet.
 			// The error must be handled here so the log entry will record the response size.
 			if l.ErrorFunc != nil {
-				l.ErrorFunc(rrw, r, rc)
+				l.ErrorFunc(ctx, rrw, r, rc)
 			} else {
 				answer := new(dns.Msg)
 				answer.SetRcode(r, rc)
 				state.SizeAndDo(answer)
 
-				vars.Report(state, vars.Dropped, rcode.ToString(rc), answer.Len(), time.Now())
+				vars.Report(ctx, state, vars.Dropped, rcode.ToString(rc), answer.Len(), time.Now())
 
 				w.WriteMsg(answer)
 			}

plugin/metrics/README.md

@@ -11,18 +11,21 @@ The default location for the metrics is `localhost:9153`. The metrics path is fi
 The following metrics are exported:
 
 * `coredns_build_info{version, revision, goversion}` - info about CoreDNS itself.
-* `coredns_dns_request_count_total{zone, proto, family}` - total query count.
-* `coredns_dns_request_duration_seconds{zone}` - duration to process each query.
-* `coredns_dns_request_size_bytes{zone, proto}` - size of the request in bytes.
-* `coredns_dns_request_do_count_total{zone}` -  queries that have the DO bit set
-* `coredns_dns_request_type_count_total{zone, type}` - counter of queries per zone and type.
-* `coredns_dns_response_size_bytes{zone, proto}` - response size in bytes.
-* `coredns_dns_response_rcode_count_total{zone, rcode}` - response per zone and rcode.
+* `coredns_dns_request_count_total{server, zone, proto, family}` - total query count.
+* `coredns_dns_request_duration_seconds{server, zone}` - duration to process each query.
+* `coredns_dns_request_size_bytes{server, zone, proto}` - size of the request in bytes.
+* `coredns_dns_request_do_count_total{server, zone}` -  queries that have the DO bit set
+* `coredns_dns_request_type_count_total{server, zone, type}` - counter of queries per zone and type.
+* `coredns_dns_response_size_bytes{server, zone, proto}` - response size in bytes.
+* `coredns_dns_response_rcode_count_total{server, zone, rcode}` - response per zone and rcode.
 
 Each counter has a label `zone` which is the zonename used for the request/response.
 
 Extra labels used are:
 
+* `server` is identifying the server responsible for the request. This is a string formatted
+  as the server's listening address: `<scheme>://[<bind>]:<port>`. I.e. for a "normal" DNS server
+  this is `dns://:53`. If you are using the *bind* plugin an IP address is included, e.g.: `dns://127.0.0.53:53`.
 * `proto` which holds the transport of the response ("udp" or "tcp")
 * The address family (`family`) of the transport (1 = IP (IP version 4), 2 = IP6 (IP version 6)).
 * `type` which holds the query type. It holds most common types (A, AAAA, MX, SOA, CNAME, PTR, TXT,

plugin/metrics/context.go

 package metrics
 
 import (
-	"github.com/coredns/coredns/plugin"
+	"github.com/coredns/coredns/plugin/metrics/vars"
 
 	"golang.org/x/net/context"
 )
@@ -15,10 +15,4 @@ import (
 // Basic usage with a metric:
 //
 // <metric>.WithLabelValues(metrics.WithServer(ctx), labels..).Add(1)
-func WithServer(ctx context.Context) string {
-	srv := ctx.Value(plugin.ServerCtx{})
-	if srv == nil {
-		return ""
-	}
-	return srv.(string)
-}
+func WithServer(ctx context.Context) string { return vars.WithServer(ctx) }

plugin/metrics/handler.go

@@ -25,7 +25,7 @@ func (m *Metrics) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg
 	rw := dnstest.NewRecorder(w)
 	status, err := plugin.NextOrFailure(m.Name(), m.Next, ctx, rw, r)
 
-	vars.Report(state, zone, rcode.ToString(rw.Rcode), rw.Len, rw.Start)
+	vars.Report(ctx, state, zone, rcode.ToString(rw.Rcode), rw.Len, rw.Start)
 
 	return status, err
 }

plugin/metrics/vars/report.go

@@ -3,13 +3,15 @@ package vars
 import (
 	"time"
 
+	"github.com/coredns/coredns/plugin"
 	"github.com/coredns/coredns/request"
 
 	"github.com/miekg/dns"
+	"golang.org/x/net/context"
 )
 
 // Report reports the metrics data associcated with request.
-func Report(req request.Request, zone, rcode string, size int, start time.Time) {
+func Report(ctx context.Context, req request.Request, zone, rcode string, size int, start time.Time) {
 	// Proto and Family.
 	net := req.Proto()
 	fam := "1"
@@ -17,25 +19,35 @@ func Report(req request.Request, zone, rcode string, size int, start time.Time)
 		fam = "2"
 	}
 
-	typ := req.QType()
+	server := WithServer(ctx)
 
-	RequestCount.WithLabelValues(zone, net, fam).Inc()
-	RequestDuration.WithLabelValues(zone).Observe(time.Since(start).Seconds())
+	typ := req.QType()
+	RequestCount.WithLabelValues(server, zone, net, fam).Inc()
+	RequestDuration.WithLabelValues(server, zone).Observe(time.Since(start).Seconds())
 
 	if req.Do() {
-		RequestDo.WithLabelValues(zone).Inc()
+		RequestDo.WithLabelValues(server, zone).Inc()
 	}
 
 	if _, known := monitorType[typ]; known {
-		RequestType.WithLabelValues(zone, dns.Type(typ).String()).Inc()
+		RequestType.WithLabelValues(server, zone, dns.Type(typ).String()).Inc()
 	} else {
-		RequestType.WithLabelValues(zone, other).Inc()
+		RequestType.WithLabelValues(server, zone, other).Inc()
 	}
 
-	ResponseSize.WithLabelValues(zone, net).Observe(float64(size))
-	RequestSize.WithLabelValues(zone, net).Observe(float64(req.Len()))
+	ResponseSize.WithLabelValues(server, zone, net).Observe(float64(size))
+	RequestSize.WithLabelValues(server, zone, net).Observe(float64(req.Len()))
 
-	ResponseRcode.WithLabelValues(zone, rcode).Inc()
+	ResponseRcode.WithLabelValues(server, zone, rcode).Inc()
+}
+
+// WithServer returns the current server handling the request.
+func WithServer(ctx context.Context) string {
+	srv := ctx.Value(plugin.ServerCtx{})
+	if srv == nil {
+		return ""
+	}
+	return srv.(string)
 }
 
 var monitorType = map[uint16]bool{

plugin/metrics/vars/vars.go

@@ -13,7 +13,7 @@ var (
 		Subsystem: subsystem,
 		Name:      "request_count_total",
 		Help:      "Counter of DNS requests made per zone, protocol and family.",
-	}, []string{"zone", "proto", "family"})
+	}, []string{"server", "zone", "proto", "family"})
 
 	RequestDuration = prometheus.NewHistogramVec(prometheus.HistogramOpts{
 		Namespace: plugin.Namespace,
@@ -21,7 +21,7 @@ var (
 		Name:      "request_duration_seconds",
 		Buckets:   plugin.TimeBuckets,
 		Help:      "Histogram of the time (in seconds) each request took.",
-	}, []string{"zone"})
+	}, []string{"server", "zone"})
 
 	RequestSize = prometheus.NewHistogramVec(prometheus.HistogramOpts{
 		Namespace: plugin.Namespace,
@@ -29,21 +29,21 @@ var (
 		Name:      "request_size_bytes",
 		Help:      "Size of the EDNS0 UDP buffer in bytes (64K for TCP).",
 		Buckets:   []float64{0, 100, 200, 300, 400, 511, 1023, 2047, 4095, 8291, 16e3, 32e3, 48e3, 64e3},
-	}, []string{"zone", "proto"})
+	}, []string{"server", "zone", "proto"})
 
 	RequestDo = prometheus.NewCounterVec(prometheus.CounterOpts{
 		Namespace: plugin.Namespace,
 		Subsystem: subsystem,
 		Name:      "request_do_count_total",
 		Help:      "Counter of DNS requests with DO bit set per zone.",
-	}, []string{"zone"})
+	}, []string{"server", "zone"})
 
 	RequestType = prometheus.NewCounterVec(prometheus.CounterOpts{
 		Namespace: plugin.Namespace,
 		Subsystem: subsystem,
 		Name:      "request_type_count_total",
 		Help:      "Counter of DNS requests per type, per zone.",
-	}, []string{"zone", "type"})
+	}, []string{"server", "zone", "type"})
 
 	ResponseSize = prometheus.NewHistogramVec(prometheus.HistogramOpts{
 		Namespace: plugin.Namespace,
@@ -51,14 +51,14 @@ var (
 		Name:      "response_size_bytes",
 		Help:      "Size of the returned response in bytes.",
 		Buckets:   []float64{0, 100, 200, 300, 400, 511, 1023, 2047, 4095, 8291, 16e3, 32e3, 48e3, 64e3},
-	}, []string{"zone", "proto"})
+	}, []string{"server", "zone", "proto"})
 
 	ResponseRcode = prometheus.NewCounterVec(prometheus.CounterOpts{
 		Namespace: plugin.Namespace,
 		Subsystem: subsystem,
 		Name:      "response_rcode_count_total",
 		Help:      "Counter of response status codes.",
-	}, []string{"zone", "rcode"})
+	}, []string{"server", "zone", "rcode"})
 )
 
 const (

plugin/proxy/README.md

@@ -102,13 +102,15 @@ payload over HTTPS). Note that with `https_google` the entire transport is encry
 
 If monitoring is enabled (via the *prometheus* directive) then the following metric is exported:
 
-* `coredns_proxy_request_duration_seconds{proto, proto_proxy, family, to}` - duration per upstream
-  interaction.
-* `coredns_proxy_request_count_total{proto, proto_proxy, family, to}` - query count per upstream.
+* `coredns_proxy_request_duration_seconds{server, proto, proto_proxy, family, to}` - duration per
+  upstream interaction.
+* `coredns_proxy_request_count_total{server, proto, proto_proxy, family, to}` - query count per
+  upstream.
 
 Where `proxy_proto` is the protocol used (`dns`, `grpc`, or `https_google`) and `to` is **TO**
-specified in the config, `proto` is the protocol used by the incoming query ("tcp" or "udp").
-and family the transport family ("1" for IPv4, and "2" for IPv6).
+specified in the config, `proto` is the protocol used by the incoming query ("tcp" or "udp"), family
+the transport family ("1" for IPv4, and "2" for IPv6). `Server` is the server responsible for the
+request (and metric). See the documention in the metrics plugin.
 
 ## Examples
 

plugin/proxy/metrics.go

@@ -15,14 +15,14 @@ var (
 		Subsystem: "proxy",
 		Name:      "request_count_total",
 		Help:      "Counter of requests made per protocol, proxy protocol, family and upstream.",
-	}, []string{"proto", "proxy_proto", "family", "to"})
+	}, []string{"server", "proto", "proxy_proto", "family", "to"})
 	RequestDuration = prometheus.NewHistogramVec(prometheus.HistogramOpts{
 		Namespace: plugin.Namespace,
 		Subsystem: "proxy",
 		Name:      "request_duration_seconds",
 		Buckets:   plugin.TimeBuckets,
 		Help:      "Histogram of the time (in seconds) each request took.",
-	}, []string{"proto", "proxy_proto", "family", "to"})
+	}, []string{"server", "proto", "proxy_proto", "family", "to"})
 )
 
 // familyToString returns the string form of either 1, or 2. Returns

plugin/proxy/proxy.go

@@ -9,6 +9,7 @@ import (
 	"time"
 
 	"github.com/coredns/coredns/plugin"
+	"github.com/coredns/coredns/plugin/metrics"
 	"github.com/coredns/coredns/plugin/pkg/healthcheck"
 	"github.com/coredns/coredns/request"
 
@@ -87,7 +88,7 @@ func (p Proxy) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg) (
 
 			atomic.AddInt64(&host.Conns, 1)
 
-			RequestCount.WithLabelValues(state.Proto(), upstream.Exchanger().Protocol(), familyToString(state.Family()), host.Name).Add(1)
+			RequestCount.WithLabelValues(metrics.WithServer(ctx), state.Proto(), upstream.Exchanger().Protocol(), familyToString(state.Family()), host.Name).Add(1)
 
 			reply, backendErr = upstream.Exchanger().Exchange(ctx, host.Name, state)
 
@@ -110,7 +111,7 @@ func (p Proxy) ServeDNS(ctx context.Context, w dns.ResponseWriter, r *dns.Msg) (
 
 				w.WriteMsg(reply)
 
-				RequestDuration.WithLabelValues(state.Proto(), upstream.Exchanger().Protocol(), familyToString(state.Family()), host.Name).Observe(time.Since(start).Seconds())
+				RequestDuration.WithLabelValues(metrics.WithServer(ctx), state.Proto(), upstream.Exchanger().Protocol(), familyToString(state.Family()), host.Name).Observe(time.Since(start).Seconds())
 
 				return 0, taperr
 			}