prober: add address family lable for udp metrics

Add a label which differentiates the address family
for STUN checks.

Also initialize the derpprobe_attempts_total and
derpprobe_seconds_total metrics by adding 0 for
the alternate fail/ok case.

Updates tailscale/corp#27249

Signed-off-by: Mike O'Driscoll <mikeo@tailscale.com>

prober/derp.go

@@ -157,7 +157,7 @@ func DERP(p *Prober, derpMapURL string, opts ...DERPOpt) (*derpProber, error) {
 	}
 	d.ProbeMap = ProbeClass{
 		Probe: d.probeMapFn,
-		Class: "derp_map",
+		Class: ClassDERPMap,
 	}
 	for _, o := range opts {
 		o(d)
@@ -286,7 +286,7 @@ func (d *derpProber) probeMesh(from, to string) ProbeClass {
 			dm := d.lastDERPMap
 			return derpProbeNodePair(ctx, dm, fromN, toN)
 		},
-		Class:  "derp_mesh",
+		Class:  ClassDERPMesh,
 		Labels: Labels{"derp_path": derpPath},
 	}
 }
@@ -310,7 +310,7 @@ func (d *derpProber) probeBandwidth(from, to string, size int64) ProbeClass {
 			}
 			return derpProbeBandwidth(ctx, d.lastDERPMap, fromN, toN, size, &transferTimeSeconds, &totalBytesTransferred, d.bwTUNIPv4Prefix)
 		},
-		Class: "derp_bw",
+		Class: ClassBandwidth,
 		Labels: Labels{
 			"derp_path":  derpPath,
 			"tcp_in_tcp": strconv.FormatBool(d.bwTUNIPv4Prefix != nil),
@@ -351,7 +351,7 @@ func (d *derpProber) probeQueuingDelay(from, to string, packetsPerSecond int, pa
 			}
 			return derpProbeQueuingDelay(ctx, d.lastDERPMap, fromN, toN, packetsPerSecond, packetTimeout, &packetsDropped, qdh)
 		},
-		Class:  "derp_qd",
+		Class:  ClassQueueDepth,
 		Labels: Labels{"derp_path": derpPath},
 		Metrics: func(l prometheus.Labels) []prometheus.Metric {
 			qdh.mx.Lock()
@@ -600,7 +600,7 @@ func (d *derpProber) ProbeUDP(ipaddr string, port int) ProbeClass {
 		Probe: func(ctx context.Context) error {
 			return derpProbeUDP(ctx, ipaddr, port)
 		},
-		Class: "derp_udp",
+		Class: ClassDERPUDP,
 	}
 }
 

prober/prober.go

@@ -17,6 +17,7 @@ import (
 	"maps"
 	"math/rand"
 	"net/http"
+	"strings"
 	"sync"
 	"time"
 
@@ -29,6 +30,17 @@ import (
 // in memory.
 const recentHistSize = 10
 
+type ClassType string
+
+var (
+	ClassDERPUDP    ClassType = "derp_udp"
+	ClassDERPMesh   ClassType = "derp_mesh"
+	ClassDERPMap    ClassType = "derp_map"
+	ClassTLS        ClassType = "tls"
+	ClassQueueDepth ClassType = "derp_qd"
+	ClassBandwidth  ClassType = "derp_bw"
+)
+
 // ProbeClass defines a probe of a specific type: a probing function that will
 // be regularly ran, and metric labels that will be added automatically to all
 // probes using this class.
@@ -40,7 +52,7 @@ type ProbeClass struct {
 
 	// Class defines a user-facing name of the probe class that will be used
 	// in the `class` metric label.
-	Class string
+	Class ClassType
 
 	// Labels defines a set of metric labels that will be added to all metrics
 	// exposed by this probe class.
@@ -117,7 +129,14 @@ func (p *Prober) Run(name string, interval time.Duration, labels Labels, pc Prob
 
 	l := prometheus.Labels{
 		"name":  name,
-		"class": pc.Class,
+		"class": string(pc.Class),
+	}
+	if pc.Class == ClassDERPUDP {
+		if strings.Contains(name, "udp6") {
+			l["af"] = "udp6"
+		} else {
+			l["af"] = "udp"
+		}
 	}
 	for k, v := range pc.Labels {
 		l[k] = v
@@ -404,10 +423,14 @@ func (p *Probe) recordEndLocked(err error) {
 		p.mSeconds.WithLabelValues("ok").Add(latency.Seconds())
 		p.latencyHist.Value = latency
 		p.latencyHist = p.latencyHist.Next()
+		p.mAttempts.WithLabelValues("fail").Add(0)
+		p.mSeconds.WithLabelValues("fail").Add(0)
 	} else {
 		p.latency = 0
 		p.mAttempts.WithLabelValues("fail").Inc()
 		p.mSeconds.WithLabelValues("fail").Add(latency.Seconds())
+		p.mAttempts.WithLabelValues("ok").Add(0)
+		p.mSeconds.WithLabelValues("ok").Add(0)
 	}
 	p.successHist.Value = p.succeeded
 	p.successHist = p.successHist.Next()
@@ -486,7 +509,7 @@ func (p *Prober) ProbeInfo() map[string]ProbeInfo {
 func (probe *Probe) probeInfoLocked() ProbeInfo {
 	inf := ProbeInfo{
 		Name:     probe.name,
-		Class:    probe.probeClass.Class,
+		Class:    string(probe.probeClass.Class),
 		Interval: probe.interval,
 		Labels:   probe.metricLabels,
 		Start:    probe.start,

prober/tls.go

@@ -36,7 +36,7 @@ func TLS(hostPort string) ProbeClass {
 			}
 			return probeTLS(ctx, certDomain, hostPort)
 		},
-		Class: "tls",
+		Class: ClassTLS,
 	}
 }
 
@@ -48,7 +48,7 @@ func TLSWithIP(certDomain string, dialAddr netip.AddrPort) ProbeClass {
 		Probe: func(ctx context.Context) error {
 			return probeTLS(ctx, certDomain, dialAddr.String())
 		},
-		Class: "tls",
+		Class: ClassTLS,
 	}
 }