feat: metrics for services and checks

client/identities.go

@@ -26,9 +26,9 @@ type Identity struct {
 	Access IdentityAccess `json:"access" yaml:"access"`
 
 	// One or more of the following type-specific configuration fields must be
-	// non-nil (currently the only types are "local" and "basicauth").
-	Local     *LocalIdentity     `json:"local,omitempty" yaml:"local,omitempty"`
-	BasicAuth *BasicAuthIdentity `json:"basicauth,omitempty" yaml:"basicauth,omitempty"`
+	// non-nil (currently the only types are "local" and "basic").
+	Local *LocalIdentity `json:"local,omitempty" yaml:"local,omitempty"`
+	Basic *BasicIdentity `json:"basic,omitempty" yaml:"basic,omitempty"`
 }
 
 // IdentityAccess defines the access level for an identity.
@@ -48,10 +48,9 @@ type LocalIdentity struct {
 	UserID *uint32 `json:"user-id" yaml:"user-id"`
 }
 
-// BasicAuthIdentity holds identity configuration specific to the "basicauth" type
+// BasicIdentity holds identity configuration specific to the "basic" type
 // (for username/password authentication).
-type BasicAuthIdentity struct {
-	Username string `json:"username" yaml:"username"`
+type BasicIdentity struct {
 	Password string `json:"password" yaml:"password"`
 }
 

internals/cli/cmd_identities.go

@@ -112,8 +112,8 @@ func (cmd *cmdIdentities) writeText(identities map[string]*client.Identity) erro
 		if identity.Local != nil {
 			types = append(types, "local")
 		}
-		if identity.BasicAuth != nil {
-			types = append(types, "basicauth")
+		if identity.Basic != nil {
+			types = append(types, "basic")
 		}
 		sort.Strings(types)
 		if len(types) == 0 {

internals/cli/cmd_run.go

@@ -248,24 +248,6 @@ func runDaemon(rcmd *cmdRun, ch chan os.Signal, ready chan<- func()) error {
 		}
 	}
 
-	metricsEndpointUsername := os.Getenv("METRICS_ENDPOINT_USERNAME")
-	metricsEndpointPassword := os.Getenv("METRICS_ENDPOINT_PASSWORD")
-	if metricsEndpointUsername != "" && metricsEndpointPassword != "" {
-		identities := map[string]*client.Identity{
-			metricsEndpointUsername: &client.Identity{
-				Access: client.ReadAccess,
-				BasicAuth: &client.BasicAuthIdentity{
-					Username: metricsEndpointUsername,
-					Password: metricsEndpointPassword,
-				},
-			},
-		}
-		err = rcmd.client.ReplaceIdentities(identities)
-		if err != nil {
-			return fmt.Errorf("cannot replace identities: %w", err)
-		}
-	}
-
 	// The "stop" channel is used by the "enter" command to stop the daemon.
 	var stop chan struct{}
 	if ready != nil {

internals/daemon/access.go

@@ -68,3 +68,18 @@ func (ac UserAccess) CheckAccess(d *Daemon, r *http.Request, user *UserState) Re
 	// An identity explicitly set to "access: untrusted" isn't allowed.
 	return Unauthorized(accessDenied)
 }
+
+// MetricsAccess allows requests over the UNIX domain socket from any local user
+type MetricsAccess struct{}
+
+func (ac MetricsAccess) CheckAccess(d *Daemon, r *http.Request, user *UserState) Response {
+	if user == nil {
+		return Unauthorized(accessDenied)
+	}
+	switch user.Identity.Access {
+	case state.MetricsAccess, state.AdminAccess:
+		return nil
+	}
+	// An identity explicitly set to "access: untrusted" isn't allowed.
+	return Unauthorized(accessDenied)
+}

internals/daemon/api.go

@@ -110,7 +110,7 @@ var API = []*Command{{
 	POST:        v1PostIdentities,
 }, {
 	Path:       "/metrics",
-	ReadAccess: UserAccess{},
+	ReadAccess: MetricsAccess{},
 	GET:        Metrics,
 }}
 

internals/metrics/metrics.go

@@ -81,7 +81,12 @@ func formatLabelKey(labels []string, labelValues []string) string {
 
 	// Sort labels for consistency
 	sort.Strings(labelPairs)
-	return strings.Join(labelPairs, ",")
+	res := strings.Join(labelPairs, ",")
+	if res == "" {
+		// a special key for situations where no labels are used
+		res = "__empty__"
+	}
+	return res
 }
 
 // NewCounterVec creates a new counter vector.
@@ -117,8 +122,8 @@ func (r *MetricsRegistry) newMetricVec(name, help string, labels []string, metri
 func (v *MetricVec) WithLabelValues(labelValues ...string) *Metric {
 	if len(labelValues) != len(v.labels) {
 		panic(fmt.Errorf(
-			"%q has %d variable labels named %q but %d values %q were provided",
-			v,
+			"%s has %d variable labels named %q but %d values %q were provided",
+			v.Name,
 			len(v.labels),
 			v.labels,
 			len(labelValues),
@@ -192,15 +197,29 @@ func (r *MetricsRegistry) GatherMetrics() string {
 		output += fmt.Sprintf("# HELP %s %s\n", vec.Name, vec.Help)
 		output += fmt.Sprintf("# TYPE %s %s\n", vec.Name, vec.Type)
 
-		for labelKey, metric := range vec.metrics {
-			switch v := metric.value.(type) {
-			case int64:
-				output += fmt.Sprintf("%s{%s} %d\n", vec.Name, labelKey, v)
-			case float64:
-				output += fmt.Sprintf("%s{%s} %f\n", vec.Name, labelKey, v)
-			default:
-				// Fallback for other types
-				output += fmt.Sprintf("%s{%s} %v\n", vec.Name, labelKey, v)
+		if len(vec.labels) == 0 { // Handle metrics without labels
+			for _, metric := range vec.metrics {
+				switch v := metric.value.(type) {
+				case int64:
+					output += fmt.Sprintf("%s %d\n", vec.Name, v) // No curly braces
+				case float64:
+					output += fmt.Sprintf("%s %.2f\n", vec.Name, v) // No curly braces
+				default:
+					output += fmt.Sprintf("%s %v\n", vec.Name, v) // Fallback
+				}
+			}
+		} else {
+			// Handle metrics with labels.
+			for labelKey, metric := range vec.metrics {
+				switch v := metric.value.(type) {
+				case int64:
+					output += fmt.Sprintf("%s{%s} %d\n", vec.Name, labelKey, v)
+				case float64:
+					output += fmt.Sprintf("%s{%s} %.2f\n", vec.Name, labelKey, v)
+				default:
+					// Fallback for other types
+					output += fmt.Sprintf("%s{%s} %v\n", vec.Name, labelKey, v)
+				}
 			}
 		}
 	}

internals/metrics/metrics_test.go

@@ -33,53 +33,66 @@ type RegistryTestSuite struct {
 
 func (s *RegistryTestSuite) SetUpTest(c *C) {
 	s.registry = &MetricsRegistry{
-		metrics: make(map[string]*Metric),
+		metricVecs: make(map[string]*MetricVec),
 	}
 }
 
-func (s *RegistryTestSuite) TestCounter(c *C) {
-	s.registry.NewMetric("test_counter", MetricTypeCounter, "Test counter")
-	s.registry.IncCounter("test_counter")
-	s.registry.IncCounter("test_counter")
-	c.Check(s.registry.metrics["test_counter"].value.(int64), Equals, int64(2))
+func (s *RegistryTestSuite) TestCounterWithoutLabels(c *C) {
+	labels := []string{}
+	testCounter := s.registry.NewCounterVec("test_counter", "Total number of something processed", labels)
+	testCounter.WithLabelValues().Inc()
+	c.Check(s.registry.metricVecs["test_counter"].metrics[formatLabelKey(labels, []string{})].value.(int64), Equals, int64(1))
+	testCounter.WithLabelValues().Inc()
+	c.Check(s.registry.metricVecs["test_counter"].metrics[formatLabelKey(labels, []string{})].value.(int64), Equals, int64(2))
 }
 
-func (s *RegistryTestSuite) TestGauge(c *C) {
-	s.registry.NewMetric("test_gauge", MetricTypeGauge, "Test gauge")
-	s.registry.SetGauge("test_gauge", 10)
-	c.Check(s.registry.metrics["test_gauge"].value.(int64), Equals, int64(10))
-	s.registry.SetGauge("test_gauge", 20)
-	c.Check(s.registry.metrics["test_gauge"].value.(int64), Equals, int64(20))
+func (s *RegistryTestSuite) TestCounterWithLabels(c *C) {
+	labels := []string{"operation", "status"}
+	testCounter := s.registry.NewCounterVec("test_counter", "Total number of something processed", labels)
+	testCounter.WithLabelValues("read", "success").Inc()
+	c.Check(s.registry.metricVecs["test_counter"].metrics[formatLabelKey(labels, []string{"read", "success"})].value.(int64), Equals, int64(1))
+	testCounter.WithLabelValues("write", "fail").Add(2)
+	c.Check(s.registry.metricVecs["test_counter"].metrics[formatLabelKey(labels, []string{"write", "fail"})].value.(int64), Equals, int64(2))
 }
 
-func (s *RegistryTestSuite) TestHistogram(c *C) {
-	s.registry.NewMetric("test_histogram", MetricTypeHistogram, "Test histogram")
-	s.registry.ObserveHistogram("test_histogram", 1.0)
-	s.registry.ObserveHistogram("test_histogram", 2.0)
-	histogramValues := s.registry.metrics["test_histogram"].value.([]float64)
-	c.Check(len(histogramValues), Equals, 2)
-	c.Check(histogramValues[0], Equals, 1.0)
-	c.Check(histogramValues[1], Equals, 2.0)
+func (s *RegistryTestSuite) TestGauge(c *C) {
+	labels := []string{"sensor"}
+	testGauge := s.registry.NewGaugeVec("test_gauge", "Current value of something", labels)
+	testGauge.WithLabelValues("temperature").Set(10.0)
+	c.Check(s.registry.metricVecs["test_gauge"].metrics[formatLabelKey(labels, []string{"temperature"})].value.(float64), Equals, float64(10.0))
+	testGauge.WithLabelValues("temperature").Set(20.0)
+	c.Check(s.registry.metricVecs["test_gauge"].metrics[formatLabelKey(labels, []string{"temperature"})].value.(float64), Equals, float64(20.0))
 }
 
 func (s *RegistryTestSuite) TestGatherMetrics(c *C) {
-	s.registry.NewMetric("test_counter", MetricTypeCounter, "Test counter")
-	s.registry.IncCounter("test_counter")
+	testCounter := s.registry.NewCounterVec("test_counter", "Total number of something processed", []string{"operation", "status"})
+	testCounter.WithLabelValues("read", "success").Inc()
+	testGauge := s.registry.NewGaugeVec("test_gauge", "Current value of something", []string{"sensor"})
+	testGauge.WithLabelValues("temperature").Set(10.0)
+	metricsOutput := s.registry.GatherMetrics()
+	expectedOutput := "# HELP test_counter Total number of something processed\n# TYPE test_counter counter\ntest_counter{operation=read,status=success} 1\n"
+	expectedOutput += "# HELP test_gauge Current value of something\n# TYPE test_gauge gauge\ntest_gauge{sensor=temperature} 10.00\n"
+	c.Check(metricsOutput, Equals, expectedOutput)
+}
+
+func (s *RegistryTestSuite) TestGatherMetricsWithoutLabels(c *C) {
+	testCounter := s.registry.NewCounterVec("test_counter", "Total number of something processed", []string{})
+	testCounter.WithLabelValues().Inc()
 	metricsOutput := s.registry.GatherMetrics()
-	expectedOutput := "# HELP test_counter Test counter\n# TYPE test_counter counter\ntest_counter 1\n"
+	expectedOutput := "# HELP test_counter Total number of something processed\n# TYPE test_counter counter\ntest_counter 1\n"
 	c.Check(metricsOutput, Equals, expectedOutput)
 }
 
 func (s *RegistryTestSuite) TestRaceConditions(c *C) {
-	s.registry.NewMetric("race_counter", MetricTypeCounter, "Race counter")
+	counter := s.registry.NewCounterVec("test_counter", "Total number of something processed", []string{})
 	var wg sync.WaitGroup
 	for i := 0; i < 1000; i++ {
 		wg.Add(1)
 		go func() {
 			defer wg.Done()
-			s.registry.IncCounter("race_counter")
+			counter.WithLabelValues().Inc()
 		}()
 	}
 	wg.Wait()
-	c.Check(s.registry.metrics["race_counter"].value.(int64), Equals, int64(1000))
+	c.Check(s.registry.metricVecs["test_counter"].metrics[formatLabelKey([]string{}, []string{})].value.(int64), Equals, int64(1000))
 }