Добавлены метрики Prometheus и дашборды Grafana (этап 2.5)

Все три Go-сервиса (ingest, device-control, rule-engine) отдают
/metrics в формате Prometheus: счётчики MQTT/RabbitMQ/ClickHouse
операций, гистограммы длительности батч-флашей и диспатча правил,
переходы устройств online/offline. Добавлены сервисы prometheus и
grafana в docker-compose с провижининг конфигом (datasource +
два готовых дашборда: "Ingest & Telemetry" и "Automation & Devices").

services/rule-engine-service/internal/config/config.go

@@ -5,6 +5,7 @@ package config
 import (
 	"fmt"
 	"os"
+	"strconv"
 	"time"
 )
 
@@ -16,6 +17,8 @@ type Config struct {
 	RabbitMQURL string
 
 	RuleCacheRefreshInterval time.Duration
+
+	MetricsPort int
 }
 
 func Load() (Config, error) {
@@ -47,6 +50,10 @@ func Load() (Config, error) {
 	}
 	cfg.RuleCacheRefreshInterval = interval
 
+	if cfg.MetricsPort, err = getEnvInt("RULE_ENGINE_METRICS_PORT", 9102); err != nil {
+		return Config{}, err
+	}
+
 	return cfg, nil
 }
 
@@ -57,6 +64,18 @@ func getEnv(key, fallback string) string {
 	return fallback
 }
 
+func getEnvInt(key string, fallback int) (int, error) {
+	v := os.Getenv(key)
+	if v == "" {
+		return fallback, nil
+	}
+	n, err := strconv.Atoi(v)
+	if err != nil {
+		return 0, fmt.Errorf("%s: %w", key, err)
+	}
+	return n, nil
+}
+
 func getEnvDuration(key string, fallback time.Duration) (time.Duration, error) {
 	v := os.Getenv(key)
 	if v == "" {

services/rule-engine-service/internal/engine/engine.go

@@ -11,6 +11,7 @@ import (
 	"time"
 
 	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/devicecontrolclient"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/metrics"
 	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
 )
 
@@ -57,6 +58,8 @@ func New(ruleSource RuleSource, dispatcher Dispatcher, publisher TriggerPublishe
 // the whole reading for. A rule the device itself rejected (bad device_id,
 // unsupported action) is terminal and does not cause a retry.
 func (e *Engine) HandleReading(ctx context.Context, r Reading) error {
+	metrics.ReadingsConsumed.Inc()
+
 	matched := e.rules.MatchingRules(r.DeviceID, r.SensorType)
 
 	var firstTransportErr error
@@ -69,8 +72,11 @@ func (e *Engine) HandleReading(ctx context.Context, r Reading) error {
 		if !matchesCondition {
 			continue
 		}
+		metrics.RulesMatched.Inc()
 
+		start := time.Now()
 		result, dispatchErr := e.dispatcher.Dispatch(ctx, rule.TargetDeviceID, rule.ActionType, rule.ActionParams)
+		metrics.DispatchDuration.WithLabelValues(rule.ActionType).Observe(time.Since(start).Seconds())
 
 		success := dispatchErr == nil && result.Success
 		errMsg := result.Error
@@ -78,13 +84,16 @@ func (e *Engine) HandleReading(ctx context.Context, r Reading) error {
 		case dispatchErr != nil:
 			errMsg = dispatchErr.Error()
 			e.logger.Error("dispatch action failed", "rule_id", rule.ID, "target_device_id", rule.TargetDeviceID, "error", dispatchErr)
+			metrics.RulesTriggered.WithLabelValues(rule.ActionType, "transport_error").Inc()
 			if firstTransportErr == nil {
 				firstTransportErr = dispatchErr
 			}
 		case !result.Success:
 			e.logger.Warn("device rejected command", "rule_id", rule.ID, "target_device_id", rule.TargetDeviceID, "error", result.Error)
+			metrics.RulesTriggered.WithLabelValues(rule.ActionType, "rejected").Inc()
 		default:
 			e.logger.Info("rule triggered", "rule_id", rule.ID, "target_device_id", rule.TargetDeviceID, "action_type", rule.ActionType)
+			metrics.RulesTriggered.WithLabelValues(rule.ActionType, "success").Inc()
 		}
 
 		if pubErr := e.publisher.PublishRuleTriggered(ctx, rule.ID, rule.ZoneID, rule.TargetDeviceID, rule.ActionType, success, errMsg); pubErr != nil {

services/rule-engine-service/internal/engine/engine_test.go

@@ -7,7 +7,10 @@ import (
 	"log/slog"
 	"testing"
 
+	"github.com/prometheus/client_golang/prometheus/testutil"
+
 	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/devicecontrolclient"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/metrics"
 	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
 )
 
@@ -154,3 +157,30 @@ func TestHandleReading_InvalidOperatorSkipsRuleButContinues(t *testing.T) {
 		t.Fatalf("got dispatch calls %+v, want only the valid rule dispatched", dispatcher.calls)
 	}
 }
+
+func TestHandleReading_RecordsMetrics(t *testing.T) {
+	source := &fakeRuleSource{rules: []rules.Rule{
+		{ID: 1, ConditionOperator: ">", ConditionValue: 28, TargetDeviceID: "fan-1", ActionType: "turn_on"},
+	}}
+	dispatcher := &fakeDispatcher{result: devicecontrolclient.Result{Success: true}}
+	e := New(source, dispatcher, &fakePublisher{}, discardLogger())
+
+	triggeredCounter := metrics.RulesTriggered.WithLabelValues("turn_on", "success")
+	beforeConsumed := testutil.ToFloat64(metrics.ReadingsConsumed)
+	beforeMatched := testutil.ToFloat64(metrics.RulesMatched)
+	beforeTriggered := testutil.ToFloat64(triggeredCounter)
+
+	if err := e.HandleReading(context.Background(), Reading{Value: 30}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	if got := testutil.ToFloat64(metrics.ReadingsConsumed); got != beforeConsumed+1 {
+		t.Fatalf("got readings_consumed %v, want %v", got, beforeConsumed+1)
+	}
+	if got := testutil.ToFloat64(metrics.RulesMatched); got != beforeMatched+1 {
+		t.Fatalf("got rules_matched %v, want %v", got, beforeMatched+1)
+	}
+	if got := testutil.ToFloat64(triggeredCounter); got != beforeTriggered+1 {
+		t.Fatalf("got rules_triggered{turn_on,success} %v, want %v", got, beforeTriggered+1)
+	}
+}

services/rule-engine-service/internal/metrics/metrics.go

@@ -0,0 +1,37 @@
+// Package metrics defines rule-engine-service's Prometheus metrics.
+// Registered automatically (via promauto) into the default registry on
+// import; served by main.go's dedicated metrics HTTP server.
+package metrics
+
+import (
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promauto"
+)
+
+var (
+	ReadingsConsumed = promauto.NewCounter(prometheus.CounterOpts{
+		Name: "rule_engine_readings_consumed_total",
+		Help: "Total number of telemetry.new_reading events consumed from RabbitMQ.",
+	})
+
+	RulesMatched = promauto.NewCounter(prometheus.CounterOpts{
+		Name: "rule_engine_rules_matched_total",
+		Help: "Total number of rule conditions that evaluated true and were dispatched.",
+	})
+
+	RulesTriggered = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "rule_engine_rules_triggered_total",
+		Help: "Total number of rule dispatch attempts, by action type and outcome.",
+	}, []string{"action_type", "outcome"})
+
+	DispatchDuration = promauto.NewHistogramVec(prometheus.HistogramOpts{
+		Name:    "rule_engine_dispatch_duration_seconds",
+		Help:    "Duration of the gRPC/HTTP call to device-control-service, by action type.",
+		Buckets: prometheus.DefBuckets,
+	}, []string{"action_type"})
+
+	CacheRefreshes = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "rule_engine_cache_refresh_total",
+		Help: "Total number of automation_rules cache refreshes from PostgreSQL, by outcome.",
+	}, []string{"outcome"})
+)

services/rule-engine-service/internal/rulecache/cache.go

@@ -10,6 +10,7 @@ import (
 	"sync"
 	"time"
 
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/metrics"
 	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
 )
 
@@ -88,12 +89,14 @@ func (c *Cache) loop() {
 func (c *Cache) refresh(ctx context.Context) error {
 	dbRows, err := c.fetcher.FetchActiveRules(ctx)
 	if err != nil {
+		metrics.CacheRefreshes.WithLabelValues("error").Inc()
 		return err
 	}
 	index := buildIndex(dbRows, c.logger)
 	c.mu.Lock()
 	c.index = index
 	c.mu.Unlock()
+	metrics.CacheRefreshes.WithLabelValues("success").Inc()
 	return nil
 }