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/device-control-service/internal/metrics/metrics.go

@@ -0,0 +1,32 @@
+// Package metrics defines device-control-service's Prometheus metrics.
+// Registered automatically (via promauto) into the default registry on
+// import; served alongside the command HTTP API's /metrics route.
+package metrics
+
+import (
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promauto"
+)
+
+var (
+	CommandsTotal = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "device_control_commands_total",
+		Help: "Total number of TurnOn/TurnOff/SetLevel commands dispatched, by action and outcome.",
+	}, []string{"action", "outcome"})
+
+	CommandDuration = promauto.NewHistogramVec(prometheus.HistogramOpts{
+		Name:    "device_control_command_duration_seconds",
+		Help:    "Duration of a command dispatch (Redis desired_state patch + MQTT publish), by action.",
+		Buckets: prometheus.DefBuckets,
+	}, []string{"action"})
+
+	MQTTEventsTotal = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "device_control_mqtt_events_total",
+		Help: "Total number of telemetry/ack messages received, by event type and outcome.",
+	}, []string{"event_type", "outcome"})
+
+	HealthTransitionsTotal = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "device_control_health_transitions_total",
+		Help: "Total number of device online/offline transitions detected.",
+	}, []string{"direction"})
+)

services/device-control-service/internal/server/server.go

@@ -9,8 +9,10 @@ import (
 	"encoding/json"
 	"fmt"
 	"log/slog"
+	"time"
 
 	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/metrics"
 )
 
 // ShadowPatcher is the subset of shadow.Store the server needs.
@@ -36,21 +38,21 @@ func New(shadow ShadowPatcher, mqtt CommandPublisher, logger *slog.Logger) *Serv
 }
 
 func (s *Server) TurnOn(ctx context.Context, req *devicecontrol.TurnOnRequest) (*devicecontrol.CommandResult, error) {
-	return s.dispatch(ctx, req.GetDeviceId(),
+	return s.dispatch(ctx, "turn_on", req.GetDeviceId(),
 		map[string]any{"power": "on"},
 		map[string]any{"action": "turn_on"},
 	)
 }
 
 func (s *Server) TurnOff(ctx context.Context, req *devicecontrol.TurnOffRequest) (*devicecontrol.CommandResult, error) {
-	return s.dispatch(ctx, req.GetDeviceId(),
+	return s.dispatch(ctx, "turn_off", req.GetDeviceId(),
 		map[string]any{"power": "off"},
 		map[string]any{"action": "turn_off"},
 	)
 }
 
 func (s *Server) SetLevel(ctx context.Context, req *devicecontrol.SetLevelRequest) (*devicecontrol.CommandResult, error) {
-	return s.dispatch(ctx, req.GetDeviceId(),
+	return s.dispatch(ctx, "set_level", req.GetDeviceId(),
 		map[string]any{"level": req.GetLevel()},
 		map[string]any{"action": "set_level", "level": req.GetLevel()},
 	)
@@ -60,27 +62,41 @@ func (s *Server) SetLevel(ctx context.Context, req *devicecontrol.SetLevelReques
 // gets an instant, optimistic view even if the device is offline), then hand
 // the command to MQTT. Business failures come back as CommandResult.Error,
 // not a gRPC error — the wire contract is success/error in one message.
-func (s *Server) dispatch(ctx context.Context, deviceID string, desiredPatch, commandPayload map[string]any) (*devicecontrol.CommandResult, error) {
+func (s *Server) dispatch(ctx context.Context, action, deviceID string, desiredPatch, commandPayload map[string]any) (*devicecontrol.CommandResult, error) {
+	start := time.Now()
+	result := s.doDispatch(ctx, action, deviceID, desiredPatch, commandPayload)
+
+	metrics.CommandDuration.WithLabelValues(action).Observe(time.Since(start).Seconds())
+	outcome := "success"
+	if !result.Success {
+		outcome = "error"
+	}
+	metrics.CommandsTotal.WithLabelValues(action, outcome).Inc()
+
+	return result, nil
+}
+
+func (s *Server) doDispatch(ctx context.Context, action, deviceID string, desiredPatch, commandPayload map[string]any) *devicecontrol.CommandResult {
 	if deviceID == "" {
-		return &devicecontrol.CommandResult{Success: false, Error: "device_id is required"}, nil
+		return &devicecontrol.CommandResult{Success: false, Error: "device_id is required"}
 	}
 
 	if err := s.shadow.PatchDesiredState(ctx, deviceID, desiredPatch); err != nil {
-		s.logger.Error("patch desired state failed", "device_id", deviceID, "error", err)
-		return &devicecontrol.CommandResult{Success: false, Error: "failed to update desired state"}, nil
+		s.logger.Error("patch desired state failed", "action", action, "device_id", deviceID, "error", err)
+		return &devicecontrol.CommandResult{Success: false, Error: "failed to update desired state"}
 	}
 
 	body, err := json.Marshal(commandPayload)
 	if err != nil {
-		s.logger.Error("encode command failed", "device_id", deviceID, "error", err)
-		return &devicecontrol.CommandResult{Success: false, Error: "failed to encode command"}, nil
+		s.logger.Error("encode command failed", "action", action, "device_id", deviceID, "error", err)
+		return &devicecontrol.CommandResult{Success: false, Error: "failed to encode command"}
 	}
 
 	topic := fmt.Sprintf("devices/%s/commands", deviceID)
 	if err := s.mqtt.Publish(topic, 1, false, body); err != nil {
-		s.logger.Error("publish command failed", "device_id", deviceID, "topic", topic, "error", err)
-		return &devicecontrol.CommandResult{Success: false, Error: "failed to dispatch command"}, nil
+		s.logger.Error("publish command failed", "action", action, "device_id", deviceID, "topic", topic, "error", err)
+		return &devicecontrol.CommandResult{Success: false, Error: "failed to dispatch command"}
 	}
 
-	return &devicecontrol.CommandResult{Success: true}, nil
+	return &devicecontrol.CommandResult{Success: true}
 }

services/device-control-service/internal/server/server_test.go

@@ -8,7 +8,10 @@ import (
 	"log/slog"
 	"testing"
 
+	"github.com/prometheus/client_golang/prometheus/testutil"
+
 	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/metrics"
 )
 
 type fakeShadow struct {
@@ -122,3 +125,27 @@ func TestDispatch_PublishFailure(t *testing.T) {
 		t.Fatal("expected failure when mqtt publish errors")
 	}
 }
+
+func TestDispatch_RecordsMetrics(t *testing.T) {
+	successCounter := metrics.CommandsTotal.WithLabelValues("turn_on", "success")
+	errorCounter := metrics.CommandsTotal.WithLabelValues("turn_on", "error")
+	before := testutil.ToFloat64(successCounter)
+
+	s := New(&fakeShadow{}, &fakeMQTT{}, discardLogger())
+	if _, err := s.TurnOn(context.Background(), &devicecontrol.TurnOnRequest{DeviceId: "d1"}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	if got := testutil.ToFloat64(successCounter); got != before+1 {
+		t.Fatalf("got turn_on/success counter %v, want %v", got, before+1)
+	}
+
+	beforeErr := testutil.ToFloat64(errorCounter)
+	s2 := New(&fakeShadow{err: errors.New("redis down")}, &fakeMQTT{}, discardLogger())
+	if _, err := s2.TurnOn(context.Background(), &devicecontrol.TurnOnRequest{DeviceId: "d1"}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if got := testutil.ToFloat64(errorCounter); got != beforeErr+1 {
+		t.Fatalf("got turn_on/error counter %v, want %v", got, beforeErr+1)
+	}
+}