Реализация rule-engine-service: RabbitMQ → правила из PostgreSQL → gRPC

Слушает telemetry.new_reading (ручной ack/nack, реквизишн только при
транспортных ошибках gRPC), кэширует активные automation_rules в памяти
с периодическим обновлением из PostgreSQL (JOIN с devices — резолвит
внутренние ID в external_id, которым оперируют MQTT/Redis/gRPC). Условия
правил (>,<,>=,<=,=,!=) оцениваются обобщённо, без привязки к конкретным
типам устройств. При срабатывании вызывает device-control-service по gRPC
и публикует automation.rule_triggered в RabbitMQ.

Проверено сквозным тестом через docker compose на полном пайплайне:
mosquitto_pub → ingest-service (ClickHouse + telemetry.new_reading) →
rule-engine-service (совпадение правила) → device-control-service (gRPC →
Redis desired_state + MQTT-команда) → automation.rule_triggered. Показание
ниже порога проверено отдельно — правило корректно не срабатывает.

.env.example

@@ -17,6 +17,7 @@ CLICKHOUSE_PASSWORD=change_me
 REDIS_HOST=redis
 REDIS_PORT=6379
 REDIS_PASSWORD=
+REDIS_DB=0
 
 # --- RabbitMQ (async events between Go services) ---
 RABBITMQ_HOST=rabbitmq
@@ -31,6 +32,9 @@ MQTT_PORT=1883
 
 # --- device-control-service (gRPC) ---
 DEVICE_CONTROL_GRPC_PORT=50051
+DEVICE_CONTROL_MQTT_CLIENT_ID=device-control-service
+DEVICE_CONTROL_HEALTHCHECK_TIMEOUT=60s
+DEVICE_CONTROL_HEALTHCHECK_INTERVAL=15s
 
 # --- ingest-service ---
 INGEST_MQTT_CLIENT_ID=ingest-service
@@ -39,6 +43,10 @@ INGEST_BATCH_MAX_SIZE=500
 INGEST_BATCH_FLUSH_INTERVAL=5s
 INGEST_BATCH_FLUSH_TIMEOUT=10s
 
+# --- rule-engine-service ---
+DEVICE_CONTROL_HOST=device-control-service
+RULE_ENGINE_CACHE_REFRESH_INTERVAL=15s
+
 # --- Laravel app (stage 2) ---
 APP_KEY=
 APP_URL=http://localhost:8000

.gitignore

@@ -3,6 +3,11 @@ services/**/bin/
 services/**/*.exe
 *.test
 *.out
+services/ingest-service/ingest-service
+services/device-control-service/device-control-service
+services/rule-engine-service/rule-engine-service
+services/health-check-service/health-check-service
+services/esp32-emulator/esp32-emulator
 
 # Laravel
 laravel-app/vendor/

README.md

@@ -1,31 +1,32 @@
-# Home Automation Platform
+# Платформа домашней автоматизации
 
-A microservice-based home automation platform, built as a portfolio project
+Pet-проект на микросервисной архитектуре для портфолио, демонстрирующий Go
-demonstrating Go (concurrency, message queues, MQTT), PHP/Laravel (API,
+(конкурентность, очереди сообщений, MQTT), PHP/Laravel (API, авторизация)
-authorization), and polyglot persistence (PostgreSQL, ClickHouse, Redis).
+и работу с разными БД по назначению (PostgreSQL, ClickHouse, Redis).
 
-**This is not a "growbox project."** A growbox is the first implemented
+**Это не «проект гроубокса».** Гроубокс — первая реализованная **зона**
-**zone** with a first set of **device types** (light, pump, fan, temperature
+с первым набором **типов устройств** (свет, полив, вентиляция, датчик
-sensor). The core system — device management, telemetry, automation rules,
+температуры). Ядро системы — управление устройствами, телеметрия, правила
-users, notifications — operates purely on the abstractions *zone*, *device*,
+автоматизации, пользователи, уведомления — оперирует исключительно
-*device type*, *sensor reading*, *command*, *rule*. Nothing growbox-specific
+абстракциями *зона*, *устройство*, *тип устройства*, *показание датчика*,
-is hardcoded in the core: adding a "living room" zone with a "smart plug"
+*команда*, *правило*. Ничего специфичного для гроубокса не зашито в ядре:
-device type should require zero changes to service code, only new rows in
+добавление зоны «гостиная» с типом устройства «умная розетка» не должно
+требовать изменений в коде сервисов — только новые строки в
 `device_types` / `zones` / `devices`.
 
-## Architecture
+## Архитектура
 
 ```
-ESP32 (or emulator)
+ESP32 (или эмулятор)
    │ MQTT
    ▼
-ingest-service (Go) ──► ClickHouse (raw telemetry)
+ingest-service (Go) ──► ClickHouse (сырая телеметрия)
    │
-   ▼ RabbitMQ ("new reading" event)
+   ▼ RabbitMQ (событие "новое показание")
-rule-engine-service (Go) ──► PostgreSQL (rules, cached)
+rule-engine-service (Go) ──► PostgreSQL (правила, кэшируются)
-   │ gRPC (on rule match)
+   │ gRPC (при срабатывании правила)
    ▼
-device-control-service (Go) ──► MQTT (command) ──► device
+device-control-service (Go) ──► MQTT (команда) ──► устройство
    │
    ▼
 Redis (device shadow: desired/reported state, status, last_seen)
@@ -33,68 +34,69 @@ Redis (device shadow: desired/reported state, status, last_seen)
 Laravel API (PHP) ──► PostgreSQL / ClickHouse / Redis
    │ gRPC / HTTP
    ▼
-device-control-service (manual commands from the UI)
+device-control-service (ручные команды из UI)
 ```
 
-## Stack
+## Стек
 
-| Concern | Technology |
+| Назначение | Технология |
 |---|---|
-| Device firmware / emulator | Go (`services/esp32-emulator`) |
+| Прошивка устройства / эмулятор | Go (`services/esp32-emulator`) |
-| Device ↔ server protocol | MQTT (Mosquitto) |
+| Протокол устройство ↔ сервер | MQTT (Mosquitto) |
-| Telemetry / commands / rules backend | Go |
+| Backend телеметрии / команд / правил | Go |
-| Business logic / API / dashboard | PHP (Laravel) |
+| Бизнес-логика / API / дашборд | PHP (Laravel) |
-| Users, zones, devices, rules | PostgreSQL |
+| Пользователи, зоны, устройства, правила | PostgreSQL |
-| Telemetry (time-series) | ClickHouse |
+| Телеметрия (time-series) | ClickHouse |
-| Real-time device state | Redis (Device Shadow pattern) |
+| Состояние устройств в реальном времени | Redis (паттерн Device Shadow) |
-| Inter-service async events | RabbitMQ |
+| Асинхронные события между сервисами | RabbitMQ |
-| Inter-service sync calls | gRPC / Protobuf |
+| Синхронные вызовы между сервисами | gRPC / Protobuf |
-| Metrics (stage 2.5) | Prometheus + Grafana |
+| Метрики (этап 2.5) | Prometheus + Grafana |
-| Containerization | Docker Compose |
+| Контейнеризация | Docker Compose |
-| Web frontend (stage 3) | Vue.js + Inertia.js |
+| Веб-фронтенд (этап 3) | Vue.js + Inertia.js |
-| Mobile app (stage 3) | Flutter |
+| Мобильное приложение (этап 3) | Flutter |
-| Notifications | Telegram Bot (MVP) / Firebase Cloud Messaging (later) |
+| Уведомления | Telegram-бот (MVP) / Firebase Cloud Messaging (позже) |
 
-## Layout
+## Структура репозитория
 
 ```
 services/
   ingest-service/           Go — MQTT → ClickHouse + RabbitMQ
   device-control-service/   Go — gRPC + MQTT + Redis device shadow
-  rule-engine-service/      Go — RabbitMQ consumer + rule evaluation
+  rule-engine-service/      Go — консьюмер RabbitMQ + оценка правил
-  health-check-service/     Go — offline detection ticker
+  health-check-service/     Go — тикер обнаружения офлайн-устройств
-  esp32-emulator/           Go — fake device for local dev
+  esp32-emulator/           Go — фейковое устройство для локальной разработки
-laravel-app/                PHP/Laravel — API, auth, dashboard (stage 2)
+laravel-app/                PHP/Laravel — API, авторизация, дашборд (этап 2)
-proto/                      Shared gRPC contracts (device_control.proto)
+proto/                      Общие gRPC-контракты (device_control.proto)
 migrations/
   postgres/                 users, zones, devices, device_types, automation_rules
-  clickhouse/               telemetry table
+  clickhouse/               таблица telemetry
-configs/mosquitto/          MQTT broker config
+configs/mosquitto/          Конфиг MQTT-брокера
-monitoring/                 Prometheus + Grafana config (stage 2.5)
+monitoring/                 Конфиги Prometheus + Grafana (этап 2.5)
 ```
 
-Each `services/*/README.md` describes that service's responsibility and
+Каждый `services/*/README.md` описывает зону ответственности сервиса и явно
-explicit non-responsibilities before any code exists there.
+то, что в неё не входит, — ещё до того, как там появится код.
 
-## Build stages
+## Этапы разработки
 
-1. **Go services** — ingest, rule-engine, device-control, ESP32 emulator.
+1. **Go-сервисы** — ingest, rule-engine, device-control, эмулятор ESP32.
-2. **Laravel** — auth (Sanctum, owner/viewer), CRUD for zones/devices/rules,
+2. **Laravel** — авторизация (Sanctum, owner/viewer), CRUD для зон/устройств/
-   Blade dashboard, manual device control. MVP is considered done here.
+   правил, Blade-дашборд, ручное управление устройствами. MVP считается
-3. *(optional, high value)* **Observability** — Prometheus metrics from the
+   готовым на этом этапе.
-   Go services, a Grafana dashboard or two.
+3. *(опционально, но сильно повышает ценность)* **Observability** — метрики
-4. **Extensions** (no rush) — Telegram notifications, Vue.js/Inertia
+   Prometheus с Go-сервисов, один-два дашборда в Grafana.
-   frontend, Flutter app, real ESP32 hardware.
+4. **Расширения** (не спеша) — Telegram-уведомления, фронтенд на
+   Vue.js/Inertia, приложение на Flutter, подключение реального ESP32.
 
-## Running locally
+## Запуск локально
 
-Infrastructure only for now (app services are added to `docker-compose.yml`
+Пока только инфраструктура (сервисы добавляются в `docker-compose.yml` по
-as they're implemented):
+мере реализации):
 
 ```bash
 cp .env.example .env
 docker compose up -d
 ```
 
-This brings up Mosquitto (`1883`), PostgreSQL (`5432`), ClickHouse (`8123`/`9000`),
+Поднимутся Mosquitto (`1883`), PostgreSQL (`5432`), ClickHouse (`8123`/`9000`),
-Redis (`6379`), and RabbitMQ (`5672`, management UI on `15672`).
+Redis (`6379`) и RabbitMQ (`5672`, веб-интерфейс управления на `15672`).

docker-compose.yml

@@ -116,6 +116,65 @@ services:
       - home-automation
     restart: unless-stopped
 
-  # Remaining app services (device-control-service, rule-engine-service,
+  device-control-service:
-  # health-check-service, esp32-emulator, laravel-app) are added here as they
+    build:
-  # get implemented — see services/*/README.md for the plan for each one.
+      context: .
+      dockerfile: services/device-control-service/Dockerfile
+    ports:
+      - "${DEVICE_CONTROL_GRPC_PORT}:50051"
+    environment:
+      DEVICE_CONTROL_GRPC_PORT: 50051
+      MQTT_HOST: mosquitto
+      MQTT_PORT: 1883
+      REDIS_HOST: redis
+      REDIS_PORT: 6379
+      REDIS_PASSWORD: ${REDIS_PASSWORD}
+      REDIS_DB: ${REDIS_DB}
+      RABBITMQ_HOST: rabbitmq
+      RABBITMQ_PORT: 5672
+      RABBITMQ_USER: ${RABBITMQ_USER}
+      RABBITMQ_PASSWORD: ${RABBITMQ_PASSWORD}
+      DEVICE_CONTROL_HEALTHCHECK_TIMEOUT: ${DEVICE_CONTROL_HEALTHCHECK_TIMEOUT}
+      DEVICE_CONTROL_HEALTHCHECK_INTERVAL: ${DEVICE_CONTROL_HEALTHCHECK_INTERVAL}
+    depends_on:
+      mosquitto:
+        condition: service_started
+      redis:
+        condition: service_started
+      rabbitmq:
+        condition: service_healthy
+    networks:
+      - home-automation
+    restart: unless-stopped
+
+  rule-engine-service:
+    build:
+      context: .
+      dockerfile: services/rule-engine-service/Dockerfile
+    environment:
+      POSTGRES_HOST: postgres
+      POSTGRES_PORT: 5432
+      POSTGRES_DB: ${POSTGRES_DB}
+      POSTGRES_USER: ${POSTGRES_USER}
+      POSTGRES_PASSWORD: ${POSTGRES_PASSWORD}
+      DEVICE_CONTROL_HOST: device-control-service
+      DEVICE_CONTROL_GRPC_PORT: 50051
+      RABBITMQ_HOST: rabbitmq
+      RABBITMQ_PORT: 5672
+      RABBITMQ_USER: ${RABBITMQ_USER}
+      RABBITMQ_PASSWORD: ${RABBITMQ_PASSWORD}
+      RULE_ENGINE_CACHE_REFRESH_INTERVAL: ${RULE_ENGINE_CACHE_REFRESH_INTERVAL}
+    depends_on:
+      postgres:
+        condition: service_healthy
+      rabbitmq:
+        condition: service_healthy
+      device-control-service:
+        condition: service_started
+    networks:
+      - home-automation
+    restart: unless-stopped
+
+  # Remaining app services (health-check-service, esp32-emulator,
+  # laravel-app) are added here as they get implemented — see
+  # services/*/README.md for the plan for each one.

proto/device_control/device_control.pb.go

@@ -0,0 +1,291 @@
+// Code generated by protoc-gen-go. DO NOT EDIT.
+// versions:
+// 	protoc-gen-go v1.36.11
+// 	protoc        v7.35.1
+// source: device_control/device_control.proto
+
+package device_control
+
+import (
+	protoreflect "google.golang.org/protobuf/reflect/protoreflect"
+	protoimpl "google.golang.org/protobuf/runtime/protoimpl"
+	reflect "reflect"
+	sync "sync"
+	unsafe "unsafe"
+)
+
+const (
+	// Verify that this generated code is sufficiently up-to-date.
+	_ = protoimpl.EnforceVersion(20 - protoimpl.MinVersion)
+	// Verify that runtime/protoimpl is sufficiently up-to-date.
+	_ = protoimpl.EnforceVersion(protoimpl.MaxVersion - 20)
+)
+
+type TurnOnRequest struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	DeviceId      string                 `protobuf:"bytes,1,opt,name=device_id,json=deviceId,proto3" json:"device_id,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *TurnOnRequest) Reset() {
+	*x = TurnOnRequest{}
+	mi := &file_device_control_device_control_proto_msgTypes[0]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *TurnOnRequest) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*TurnOnRequest) ProtoMessage() {}
+
+func (x *TurnOnRequest) ProtoReflect() protoreflect.Message {
+	mi := &file_device_control_device_control_proto_msgTypes[0]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use TurnOnRequest.ProtoReflect.Descriptor instead.
+func (*TurnOnRequest) Descriptor() ([]byte, []int) {
+	return file_device_control_device_control_proto_rawDescGZIP(), []int{0}
+}
+
+func (x *TurnOnRequest) GetDeviceId() string {
+	if x != nil {
+		return x.DeviceId
+	}
+	return ""
+}
+
+type TurnOffRequest struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	DeviceId      string                 `protobuf:"bytes,1,opt,name=device_id,json=deviceId,proto3" json:"device_id,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *TurnOffRequest) Reset() {
+	*x = TurnOffRequest{}
+	mi := &file_device_control_device_control_proto_msgTypes[1]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *TurnOffRequest) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*TurnOffRequest) ProtoMessage() {}
+
+func (x *TurnOffRequest) ProtoReflect() protoreflect.Message {
+	mi := &file_device_control_device_control_proto_msgTypes[1]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use TurnOffRequest.ProtoReflect.Descriptor instead.
+func (*TurnOffRequest) Descriptor() ([]byte, []int) {
+	return file_device_control_device_control_proto_rawDescGZIP(), []int{1}
+}
+
+func (x *TurnOffRequest) GetDeviceId() string {
+	if x != nil {
+		return x.DeviceId
+	}
+	return ""
+}
+
+type SetLevelRequest struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	DeviceId      string                 `protobuf:"bytes,1,opt,name=device_id,json=deviceId,proto3" json:"device_id,omitempty"`
+	Level         float64                `protobuf:"fixed64,2,opt,name=level,proto3" json:"level,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *SetLevelRequest) Reset() {
+	*x = SetLevelRequest{}
+	mi := &file_device_control_device_control_proto_msgTypes[2]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *SetLevelRequest) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*SetLevelRequest) ProtoMessage() {}
+
+func (x *SetLevelRequest) ProtoReflect() protoreflect.Message {
+	mi := &file_device_control_device_control_proto_msgTypes[2]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use SetLevelRequest.ProtoReflect.Descriptor instead.
+func (*SetLevelRequest) Descriptor() ([]byte, []int) {
+	return file_device_control_device_control_proto_rawDescGZIP(), []int{2}
+}
+
+func (x *SetLevelRequest) GetDeviceId() string {
+	if x != nil {
+		return x.DeviceId
+	}
+	return ""
+}
+
+func (x *SetLevelRequest) GetLevel() float64 {
+	if x != nil {
+		return x.Level
+	}
+	return 0
+}
+
+type CommandResult struct {
+	state         protoimpl.MessageState `protogen:"open.v1"`
+	Success       bool                   `protobuf:"varint,1,opt,name=success,proto3" json:"success,omitempty"`
+	Error         string                 `protobuf:"bytes,2,opt,name=error,proto3" json:"error,omitempty"`
+	unknownFields protoimpl.UnknownFields
+	sizeCache     protoimpl.SizeCache
+}
+
+func (x *CommandResult) Reset() {
+	*x = CommandResult{}
+	mi := &file_device_control_device_control_proto_msgTypes[3]
+	ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+	ms.StoreMessageInfo(mi)
+}
+
+func (x *CommandResult) String() string {
+	return protoimpl.X.MessageStringOf(x)
+}
+
+func (*CommandResult) ProtoMessage() {}
+
+func (x *CommandResult) ProtoReflect() protoreflect.Message {
+	mi := &file_device_control_device_control_proto_msgTypes[3]
+	if x != nil {
+		ms := protoimpl.X.MessageStateOf(protoimpl.Pointer(x))
+		if ms.LoadMessageInfo() == nil {
+			ms.StoreMessageInfo(mi)
+		}
+		return ms
+	}
+	return mi.MessageOf(x)
+}
+
+// Deprecated: Use CommandResult.ProtoReflect.Descriptor instead.
+func (*CommandResult) Descriptor() ([]byte, []int) {
+	return file_device_control_device_control_proto_rawDescGZIP(), []int{3}
+}
+
+func (x *CommandResult) GetSuccess() bool {
+	if x != nil {
+		return x.Success
+	}
+	return false
+}
+
+func (x *CommandResult) GetError() string {
+	if x != nil {
+		return x.Error
+	}
+	return ""
+}
+
+var File_device_control_device_control_proto protoreflect.FileDescriptor
+
+const file_device_control_device_control_proto_rawDesc = "" +
+	"\n" +
+	"#device_control/device_control.proto\x12\x0edevice_control\",\n" +
+	"\rTurnOnRequest\x12\x1b\n" +
+	"\tdevice_id\x18\x01 \x01(\tR\bdeviceId\"-\n" +
+	"\x0eTurnOffRequest\x12\x1b\n" +
+	"\tdevice_id\x18\x01 \x01(\tR\bdeviceId\"D\n" +
+	"\x0fSetLevelRequest\x12\x1b\n" +
+	"\tdevice_id\x18\x01 \x01(\tR\bdeviceId\x12\x14\n" +
+	"\x05level\x18\x02 \x01(\x01R\x05level\"?\n" +
+	"\rCommandResult\x12\x18\n" +
+	"\asuccess\x18\x01 \x01(\bR\asuccess\x12\x14\n" +
+	"\x05error\x18\x02 \x01(\tR\x05error2\xed\x01\n" +
+	"\rDeviceControl\x12F\n" +
+	"\x06TurnOn\x12\x1d.device_control.TurnOnRequest\x1a\x1d.device_control.CommandResult\x12H\n" +
+	"\aTurnOff\x12\x1e.device_control.TurnOffRequest\x1a\x1d.device_control.CommandResult\x12J\n" +
+	"\bSetLevel\x12\x1f.device_control.SetLevelRequest\x1a\x1d.device_control.CommandResultB>Z<git.cactoz.su/cacto/home_automatization/proto/device_controlb\x06proto3"
+
+var (
+	file_device_control_device_control_proto_rawDescOnce sync.Once
+	file_device_control_device_control_proto_rawDescData []byte
+)
+
+func file_device_control_device_control_proto_rawDescGZIP() []byte {
+	file_device_control_device_control_proto_rawDescOnce.Do(func() {
+		file_device_control_device_control_proto_rawDescData = protoimpl.X.CompressGZIP(unsafe.Slice(unsafe.StringData(file_device_control_device_control_proto_rawDesc), len(file_device_control_device_control_proto_rawDesc)))
+	})
+	return file_device_control_device_control_proto_rawDescData
+}
+
+var file_device_control_device_control_proto_msgTypes = make([]protoimpl.MessageInfo, 4)
+var file_device_control_device_control_proto_goTypes = []any{
+	(*TurnOnRequest)(nil),   // 0: device_control.TurnOnRequest
+	(*TurnOffRequest)(nil),  // 1: device_control.TurnOffRequest
+	(*SetLevelRequest)(nil), // 2: device_control.SetLevelRequest
+	(*CommandResult)(nil),   // 3: device_control.CommandResult
+}
+var file_device_control_device_control_proto_depIdxs = []int32{
+	0, // 0: device_control.DeviceControl.TurnOn:input_type -> device_control.TurnOnRequest
+	1, // 1: device_control.DeviceControl.TurnOff:input_type -> device_control.TurnOffRequest
+	2, // 2: device_control.DeviceControl.SetLevel:input_type -> device_control.SetLevelRequest
+	3, // 3: device_control.DeviceControl.TurnOn:output_type -> device_control.CommandResult
+	3, // 4: device_control.DeviceControl.TurnOff:output_type -> device_control.CommandResult
+	3, // 5: device_control.DeviceControl.SetLevel:output_type -> device_control.CommandResult
+	3, // [3:6] is the sub-list for method output_type
+	0, // [0:3] is the sub-list for method input_type
+	0, // [0:0] is the sub-list for extension type_name
+	0, // [0:0] is the sub-list for extension extendee
+	0, // [0:0] is the sub-list for field type_name
+}
+
+func init() { file_device_control_device_control_proto_init() }
+func file_device_control_device_control_proto_init() {
+	if File_device_control_device_control_proto != nil {
+		return
+	}
+	type x struct{}
+	out := protoimpl.TypeBuilder{
+		File: protoimpl.DescBuilder{
+			GoPackagePath: reflect.TypeOf(x{}).PkgPath(),
+			RawDescriptor: unsafe.Slice(unsafe.StringData(file_device_control_device_control_proto_rawDesc), len(file_device_control_device_control_proto_rawDesc)),
+			NumEnums:      0,
+			NumMessages:   4,
+			NumExtensions: 0,
+			NumServices:   1,
+		},
+		GoTypes:           file_device_control_device_control_proto_goTypes,
+		DependencyIndexes: file_device_control_device_control_proto_depIdxs,
+		MessageInfos:      file_device_control_device_control_proto_msgTypes,
+	}.Build()
+	File_device_control_device_control_proto = out.File
+	file_device_control_device_control_proto_goTypes = nil
+	file_device_control_device_control_proto_depIdxs = nil
+}

proto/device_control/device_control.proto

@@ -0,0 +1,36 @@
+syntax = "proto3";
+
+package device_control;
+
+option go_package = "git.cactoz.su/cacto/home_automatization/proto/device_control";
+
+// DeviceControl is the synchronous entry point for issuing commands to a
+// device. It covers the generic actuator capabilities used across the
+// platform (device_types.capabilities: turn_on, turn_off, set_level) — no
+// zone- or device-type-specific methods. A call here updates the device's
+// desired state and dispatches the command over MQTT; it does not wait for
+// the device to confirm execution (see the Device Shadow pattern in
+// device-control-service's README).
+service DeviceControl {
+  rpc TurnOn(TurnOnRequest) returns (CommandResult);
+  rpc TurnOff(TurnOffRequest) returns (CommandResult);
+  rpc SetLevel(SetLevelRequest) returns (CommandResult);
+}
+
+message TurnOnRequest {
+  string device_id = 1;
+}
+
+message TurnOffRequest {
+  string device_id = 1;
+}
+
+message SetLevelRequest {
+  string device_id = 1;
+  double level = 2;
+}
+
+message CommandResult {
+  bool success = 1;
+  string error = 2;
+}

proto/device_control/device_control_grpc.pb.go

@@ -0,0 +1,213 @@
+// Code generated by protoc-gen-go-grpc. DO NOT EDIT.
+// versions:
+// - protoc-gen-go-grpc v1.6.2
+// - protoc             v7.35.1
+// source: device_control/device_control.proto
+
+package device_control
+
+import (
+	context "context"
+	grpc "google.golang.org/grpc"
+	codes "google.golang.org/grpc/codes"
+	status "google.golang.org/grpc/status"
+)
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the grpc package it is being compiled against.
+// Requires gRPC-Go v1.64.0 or later.
+const _ = grpc.SupportPackageIsVersion9
+
+const (
+	DeviceControl_TurnOn_FullMethodName   = "/device_control.DeviceControl/TurnOn"
+	DeviceControl_TurnOff_FullMethodName  = "/device_control.DeviceControl/TurnOff"
+	DeviceControl_SetLevel_FullMethodName = "/device_control.DeviceControl/SetLevel"
+)
+
+// DeviceControlClient is the client API for DeviceControl service.
+//
+// For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
+//
+// DeviceControl is the synchronous entry point for issuing commands to a
+// device. It covers the generic actuator capabilities used across the
+// platform (device_types.capabilities: turn_on, turn_off, set_level) — no
+// zone- or device-type-specific methods. A call here updates the device's
+// desired state and dispatches the command over MQTT; it does not wait for
+// the device to confirm execution (see the Device Shadow pattern in
+// device-control-service's README).
+type DeviceControlClient interface {
+	TurnOn(ctx context.Context, in *TurnOnRequest, opts ...grpc.CallOption) (*CommandResult, error)
+	TurnOff(ctx context.Context, in *TurnOffRequest, opts ...grpc.CallOption) (*CommandResult, error)
+	SetLevel(ctx context.Context, in *SetLevelRequest, opts ...grpc.CallOption) (*CommandResult, error)
+}
+
+type deviceControlClient struct {
+	cc grpc.ClientConnInterface
+}
+
+func NewDeviceControlClient(cc grpc.ClientConnInterface) DeviceControlClient {
+	return &deviceControlClient{cc}
+}
+
+func (c *deviceControlClient) TurnOn(ctx context.Context, in *TurnOnRequest, opts ...grpc.CallOption) (*CommandResult, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CommandResult)
+	err := c.cc.Invoke(ctx, DeviceControl_TurnOn_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *deviceControlClient) TurnOff(ctx context.Context, in *TurnOffRequest, opts ...grpc.CallOption) (*CommandResult, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CommandResult)
+	err := c.cc.Invoke(ctx, DeviceControl_TurnOff_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *deviceControlClient) SetLevel(ctx context.Context, in *SetLevelRequest, opts ...grpc.CallOption) (*CommandResult, error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	out := new(CommandResult)
+	err := c.cc.Invoke(ctx, DeviceControl_SetLevel_FullMethodName, in, out, cOpts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// DeviceControlServer is the server API for DeviceControl service.
+// All implementations must embed UnimplementedDeviceControlServer
+// for forward compatibility.
+//
+// DeviceControl is the synchronous entry point for issuing commands to a
+// device. It covers the generic actuator capabilities used across the
+// platform (device_types.capabilities: turn_on, turn_off, set_level) — no
+// zone- or device-type-specific methods. A call here updates the device's
+// desired state and dispatches the command over MQTT; it does not wait for
+// the device to confirm execution (see the Device Shadow pattern in
+// device-control-service's README).
+type DeviceControlServer interface {
+	TurnOn(context.Context, *TurnOnRequest) (*CommandResult, error)
+	TurnOff(context.Context, *TurnOffRequest) (*CommandResult, error)
+	SetLevel(context.Context, *SetLevelRequest) (*CommandResult, error)
+	mustEmbedUnimplementedDeviceControlServer()
+}
+
+// UnimplementedDeviceControlServer must be embedded to have
+// forward compatible implementations.
+//
+// NOTE: this should be embedded by value instead of pointer to avoid a nil
+// pointer dereference when methods are called.
+type UnimplementedDeviceControlServer struct{}
+
+func (UnimplementedDeviceControlServer) TurnOn(context.Context, *TurnOnRequest) (*CommandResult, error) {
+	return nil, status.Error(codes.Unimplemented, "method TurnOn not implemented")
+}
+func (UnimplementedDeviceControlServer) TurnOff(context.Context, *TurnOffRequest) (*CommandResult, error) {
+	return nil, status.Error(codes.Unimplemented, "method TurnOff not implemented")
+}
+func (UnimplementedDeviceControlServer) SetLevel(context.Context, *SetLevelRequest) (*CommandResult, error) {
+	return nil, status.Error(codes.Unimplemented, "method SetLevel not implemented")
+}
+func (UnimplementedDeviceControlServer) mustEmbedUnimplementedDeviceControlServer() {}
+func (UnimplementedDeviceControlServer) testEmbeddedByValue()                       {}
+
+// UnsafeDeviceControlServer may be embedded to opt out of forward compatibility for this service.
+// Use of this interface is not recommended, as added methods to DeviceControlServer will
+// result in compilation errors.
+type UnsafeDeviceControlServer interface {
+	mustEmbedUnimplementedDeviceControlServer()
+}
+
+func RegisterDeviceControlServer(s grpc.ServiceRegistrar, srv DeviceControlServer) {
+	// If the following call panics, it indicates UnimplementedDeviceControlServer was
+	// embedded by pointer and is nil.  This will cause panics if an
+	// unimplemented method is ever invoked, so we test this at initialization
+	// time to prevent it from happening at runtime later due to I/O.
+	if t, ok := srv.(interface{ testEmbeddedByValue() }); ok {
+		t.testEmbeddedByValue()
+	}
+	s.RegisterService(&DeviceControl_ServiceDesc, srv)
+}
+
+func _DeviceControl_TurnOn_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(TurnOnRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DeviceControlServer).TurnOn(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: DeviceControl_TurnOn_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DeviceControlServer).TurnOn(ctx, req.(*TurnOnRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _DeviceControl_TurnOff_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(TurnOffRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DeviceControlServer).TurnOff(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: DeviceControl_TurnOff_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DeviceControlServer).TurnOff(ctx, req.(*TurnOffRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _DeviceControl_SetLevel_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(SetLevelRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(DeviceControlServer).SetLevel(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: DeviceControl_SetLevel_FullMethodName,
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(DeviceControlServer).SetLevel(ctx, req.(*SetLevelRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+// DeviceControl_ServiceDesc is the grpc.ServiceDesc for DeviceControl service.
+// It's only intended for direct use with grpc.RegisterService,
+// and not to be introspected or modified (even as a copy)
+var DeviceControl_ServiceDesc = grpc.ServiceDesc{
+	ServiceName: "device_control.DeviceControl",
+	HandlerType: (*DeviceControlServer)(nil),
+	Methods: []grpc.MethodDesc{
+		{
+			MethodName: "TurnOn",
+			Handler:    _DeviceControl_TurnOn_Handler,
+		},
+		{
+			MethodName: "TurnOff",
+			Handler:    _DeviceControl_TurnOff_Handler,
+		},
+		{
+			MethodName: "SetLevel",
+			Handler:    _DeviceControl_SetLevel_Handler,
+		},
+	},
+	Streams:  []grpc.StreamDesc{},
+	Metadata: "device_control/device_control.proto",
+}

proto/go.mod

@@ -0,0 +1,15 @@
+module git.cactoz.su/cacto/home_automatization/proto
+
+go 1.25.0
+
+require (
+	google.golang.org/grpc v1.82.1
+	google.golang.org/protobuf v1.36.11
+)
+
+require (
+	golang.org/x/net v0.53.0 // indirect
+	golang.org/x/sys v0.43.0 // indirect
+	golang.org/x/text v0.36.0 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 // indirect
+)

proto/go.sum

@@ -0,0 +1,38 @@
+github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
+github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
+github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
+github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
+github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
+github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
+go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
+go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
+go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
+go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
+go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
+go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
+go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
+golang.org/x/net v0.53.0 h1:d+qAbo5L0orcWAr0a9JweQpjXF19LMXJE8Ey7hwOdUA=
+golang.org/x/net v0.53.0/go.mod h1:JvMuJH7rrdiCfbeHoo3fCQU24Lf5JJwT9W3sJFulfgs=
+golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
+golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/text v0.36.0 h1:JfKh3XmcRPqZPKevfXVpI1wXPTqbkE5f7JA92a55Yxg=
+golang.org/x/text v0.36.0/go.mod h1:NIdBknypM8iqVmPiuco0Dh6P5Jcdk8lJL0CUebqK164=
+gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
+gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 h1:RmoJA1ujG+/lRGNfUnOMfhCy5EipVMyvUE+KNbPbTlw=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/grpc v1.82.1 h1:NnAxzGRA0677vCa4BUkOAnO5+FfQqVl9iUXeD0IqcGE=
+google.golang.org/grpc v1.82.1/go.mod h1:yzTZ1TB1Z3SG+LIYaI+WiE8D5+PZ3ArnrSp8zF3+/ZA=
+google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
+google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=

services/device-control-service/Dockerfile

@@ -0,0 +1,15 @@
+# Build context is the repo root (not this directory) — device-control-service
+# depends on the sibling proto/ Go module via a relative `replace` directive.
+FROM golang:1.25-alpine AS build
+WORKDIR /src
+COPY proto/ ./proto/
+COPY services/device-control-service/ ./services/device-control-service/
+WORKDIR /src/services/device-control-service
+RUN go mod download
+RUN CGO_ENABLED=0 go build -o /out/device-control-service ./cmd/device-control-service
+
+FROM alpine:3.20
+RUN adduser -D -u 10001 app
+COPY --from=build /out/device-control-service /usr/local/bin/device-control-service
+USER app
+ENTRYPOINT ["/usr/local/bin/device-control-service"]

services/device-control-service/README.md

@@ -1,20 +1,51 @@
 # device-control-service (Go)
 
-Status: not implemented yet.
+Статус: реализован (MVP).
 
-Responsibility:
+Зона ответственности:
-- Expose a gRPC API (see `proto/device_control.proto`) for issuing commands
+- Предоставляет gRPC API (см. `proto/device_control/device_control.proto`:
-  to devices — called by Laravel (manual control) and rule-engine-service
+  `TurnOn`, `TurnOff`, `SetLevel` → `CommandResult`) для отправки команд
-  (rule-triggered actions).
+  устройствам — вызывается из Laravel (ручное управление) и из
-- Publish the command to the device's MQTT topic (`devices/{device_id}/commands`)
+  rule-engine-service (действия по срабатыванию правил).
-  and listen for acknowledgements (`devices/{device_id}/ack`).
+- Публикует команду в MQTT-топик устройства (`devices/{device_id}/commands`)
-- Maintain the Device Shadow pattern in Redis: `desired_state` (what the user
+  и слушает подтверждения (`devices/{device_id}/ack`).
-  wants) vs `reported_state` (what the device confirmed), plus `status` and
+- Поддерживает паттерн Device Shadow в Redis: `desired_state` (чего хочет
-  `last_seen`.
+  пользователь) против `reported_state` (что подтвердило устройство), плюс
-- Owns the health-check loop (or delegates to health-check-service): a ticker
+  `status` и `last_seen`.
-  goroutine that flips a device to `offline` when `last_seen` exceeds a
+- Ведёт health-check горутиной-тикером: переводит устройство в `offline`,
-  timeout, and emits an event for notification-service.
+  когда `last_seen` превышает таймаут, и публикует событие в RabbitMQ
+  (`device.status_changed`) для будущего notification-service — как при
+  уходе в офлайн, так и при возврате online.
 
-Not this service's job: deciding *when* to send a command based on sensor
+Не входит в зону ответственности: решение о том, *когда* отправлять команду
-thresholds — that's rule-engine-service's logic. This service only executes
+на основе показаний датчиков — это логика rule-engine-service. Этот сервис
-and tracks state for whatever command it's given.
+только исполняет команды и отслеживает состояние для той команды, что ему дали.
+
+## Архитектурные решения
+
+- **gRPC — не ждёт подтверждения устройства.** `CommandResult.success`
+  означает «desired_state обновлён и команда опубликована в MQTT», а не
+  «устройство подтвердило выполнение». Это соответствует Device Shadow:
+  мгновенный отклик в UI, даже если устройство офлайн или ответит с
+  задержкой. `reported_state` обновляется позже, асинхронно, по приходу ack.
+- **last_seen обновляется и по телеметрии, и по ack.** Сервис подписан на
+  `devices/+/telemetry` только ради отметки «устройство живо» (сами данные
+  телеметрии парсит и хранит ingest-service) — иначе датчики без исходящих
+  команд (и, соответственно, без ack) никогда не считались бы online.
+- **Нет зависимости от PostgreSQL.** Список известных устройств — это Redis
+  Set (`devices:known`), который пополняется по мере поступления
+  телеметрии/ack, а не выгружается из таблицы `devices`. Это удерживает
+  сервис в границах MQTT+Redis+RabbitMQ+gRPC, как и в диаграмме архитектуры
+  корневого README.
+
+## Запуск
+
+```bash
+cd services/device-control-service
+go test ./...
+go build ./cmd/device-control-service
+```
+
+Конфигурация — через переменные окружения (секция device-control-service в
+корневом `.env.example`). `proto/` — отдельный Go-модуль, подключается через
+`replace` в `go.mod` на относительный путь `../../proto`.

services/device-control-service/cmd/device-control-service/main.go

@@ -0,0 +1,171 @@
+// Command device-control-service exposes a gRPC API for issuing device
+// commands, dispatches them over MQTT, maintains the Device Shadow in
+// Redis, and runs a health-check loop that flips stale devices offline.
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"log/slog"
+	"net"
+	"os"
+	"os/signal"
+	"strings"
+	"syscall"
+	"time"
+
+	"google.golang.org/grpc"
+
+	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/config"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/healthcheck"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/mqttclient"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/rabbitmq"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/server"
+	"git.cactoz.su/cacto/home_automatization/services/device-control-service/internal/shadow"
+)
+
+func main() {
+	logger := slog.New(slog.NewJSONHandler(os.Stdout, nil))
+
+	if err := run(logger); err != nil {
+		logger.Error("device-control-service exited with error", "error", err)
+		os.Exit(1)
+	}
+}
+
+func run(logger *slog.Logger) error {
+	cfg, err := config.Load()
+	if err != nil {
+		return err
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	shadowStore, err := shadow.Connect(ctx, cfg.RedisAddr, cfg.RedisPassword, cfg.RedisDB)
+	cancel()
+	if err != nil {
+		return err
+	}
+	defer shadowStore.Close()
+
+	publisher, err := rabbitmq.Connect(cfg.RabbitMQURL)
+	if err != nil {
+		return err
+	}
+	defer publisher.Close()
+
+	mqttClient, err := mqttclient.Connect(mqttclient.Config{
+		BrokerURL: cfg.MQTTBrokerURL,
+		ClientID:  cfg.MQTTClientID,
+	}, logger)
+	if err != nil {
+		return err
+	}
+	defer mqttClient.Close()
+
+	if err := mqttClient.Subscribe("devices/+/telemetry", 1, telemetryHandler(shadowStore, publisher, logger)); err != nil {
+		return fmt.Errorf("subscribe telemetry: %w", err)
+	}
+	if err := mqttClient.Subscribe("devices/+/ack", 1, ackHandler(shadowStore, publisher, logger)); err != nil {
+		return fmt.Errorf("subscribe ack: %w", err)
+	}
+
+	checker := healthcheck.New(shadowStore, cfg.HealthCheckTimeout, cfg.HealthCheckInterval, func(deviceID string) {
+		pubCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+		defer cancel()
+		if err := publisher.PublishStatusChanged(pubCtx, deviceID, shadow.StatusOffline); err != nil {
+			logger.Error("publish offline event failed", "device_id", deviceID, "error", err)
+		}
+	}, logger)
+	checker.Start()
+	defer checker.Stop()
+
+	lis, err := net.Listen("tcp", fmt.Sprintf(":%d", cfg.GRPCPort))
+	if err != nil {
+		return fmt.Errorf("listen on grpc port %d: %w", cfg.GRPCPort, err)
+	}
+	grpcServer := grpc.NewServer()
+	devicecontrol.RegisterDeviceControlServer(grpcServer, server.New(shadowStore, mqttClient, logger))
+
+	go func() {
+		logger.Info("device-control-service started", "grpc_port", cfg.GRPCPort)
+		if err := grpcServer.Serve(lis); err != nil {
+			logger.Error("grpc server stopped", "error", err)
+		}
+	}()
+
+	stop := make(chan os.Signal, 1)
+	signal.Notify(stop, syscall.SIGINT, syscall.SIGTERM)
+	<-stop
+
+	logger.Info("shutting down")
+	grpcServer.GracefulStop()
+
+	return nil
+}
+
+// deviceIDFromTopic extracts {device_id} from "devices/{device_id}/<suffix>".
+func deviceIDFromTopic(topic string) (string, bool) {
+	parts := strings.Split(topic, "/")
+	if len(parts) != 3 || parts[0] != "devices" || parts[1] == "" {
+		return "", false
+	}
+	return parts[1], true
+}
+
+// telemetryHandler only tracks liveness: ingest-service owns storing and
+// interpreting telemetry, this service just needs to know the device is alive.
+func telemetryHandler(store *shadow.Store, publisher *rabbitmq.Publisher, logger *slog.Logger) mqttclient.Handler {
+	return func(topic string, _ []byte, _ time.Time) {
+		deviceID, ok := deviceIDFromTopic(topic)
+		if !ok {
+			logger.Warn("dropping telemetry from unparseable topic", "topic", topic)
+			return
+		}
+		touch(context.Background(), store, publisher, deviceID, logger)
+	}
+}
+
+type ackPayload struct {
+	State map[string]any `json:"state"`
+}
+
+func ackHandler(store *shadow.Store, publisher *rabbitmq.Publisher, logger *slog.Logger) mqttclient.Handler {
+	return func(topic string, payload []byte, _ time.Time) {
+		deviceID, ok := deviceIDFromTopic(topic)
+		if !ok {
+			logger.Warn("dropping ack from unparseable topic", "topic", topic)
+			return
+		}
+
+		var ack ackPayload
+		if err := json.Unmarshal(payload, &ack); err != nil {
+			logger.Warn("dropping invalid ack payload", "device_id", deviceID, "error", err)
+			return
+		}
+
+		ctx := context.Background()
+		if len(ack.State) > 0 {
+			if err := store.PatchReportedState(ctx, deviceID, ack.State); err != nil {
+				logger.Error("patch reported state failed", "device_id", deviceID, "error", err)
+			}
+		}
+		touch(ctx, store, publisher, deviceID, logger)
+	}
+}
+
+func touch(ctx context.Context, store *shadow.Store, publisher *rabbitmq.Publisher, deviceID string, logger *slog.Logger) {
+	becameOnline, err := store.Touch(ctx, deviceID)
+	if err != nil {
+		logger.Error("touch device failed", "device_id", deviceID, "error", err)
+		return
+	}
+	if becameOnline {
+		pubCtx, cancel := context.WithTimeout(ctx, 5*time.Second)
+		defer cancel()
+		if err := publisher.PublishStatusChanged(pubCtx, deviceID, shadow.StatusOnline); err != nil {
+			logger.Error("publish online event failed", "device_id", deviceID, "error", err)
+		}
+	}
+}

services/device-control-service/go.mod

@@ -0,0 +1,27 @@
+module git.cactoz.su/cacto/home_automatization/services/device-control-service
+
+go 1.25.0
+
+require (
+	git.cactoz.su/cacto/home_automatization/proto v0.0.0
+	github.com/alicebob/miniredis/v2 v2.38.0
+	github.com/eclipse/paho.mqtt.golang v1.5.1
+	github.com/rabbitmq/amqp091-go v1.13.0
+	github.com/redis/go-redis/v9 v9.21.0
+	google.golang.org/grpc v1.82.1
+)
+
+require (
+	github.com/cespare/xxhash/v2 v2.3.0 // indirect
+	github.com/gorilla/websocket v1.5.3 // indirect
+	github.com/yuin/gopher-lua v1.1.1 // indirect
+	go.uber.org/atomic v1.11.0 // indirect
+	golang.org/x/net v0.53.0 // indirect
+	golang.org/x/sync v0.20.0 // indirect
+	golang.org/x/sys v0.43.0 // indirect
+	golang.org/x/text v0.36.0 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 // indirect
+	google.golang.org/protobuf v1.36.11 // indirect
+)
+
+replace git.cactoz.su/cacto/home_automatization/proto => ../../proto

services/device-control-service/go.sum

@@ -0,0 +1,70 @@
+github.com/alicebob/miniredis/v2 v2.38.0 h1:nZAzCR+Lj+Vxk4ZXzm2NuKq2O33RXj1XxJ2e2uP9jiw=
+github.com/alicebob/miniredis/v2 v2.38.0/go.mod h1:TcL7YfarKPGDAthEtl5NBeHZfeUQj6OXMm/+iu5cLMM=
+github.com/bsm/ginkgo/v2 v2.12.0 h1:Ny8MWAHyOepLGlLKYmXG4IEkioBysk6GpaRTLC8zwWs=
+github.com/bsm/ginkgo/v2 v2.12.0/go.mod h1:SwYbGRRDovPVboqFv0tPTcG1sN61LM1Z4ARdbAV9g4c=
+github.com/bsm/gomega v1.27.10 h1:yeMWxP2pV2fG3FgAODIY8EiRE3dy0aeFYt4l7wh6yKA=
+github.com/bsm/gomega v1.27.10/go.mod h1:JyEr/xRbxbtgWNi8tIEVPUYZ5Dzef52k01W3YH0H+O0=
+github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
+github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/eclipse/paho.mqtt.golang v1.5.1 h1:/VSOv3oDLlpqR2Epjn1Q7b2bSTplJIeV2ISgCl2W7nE=
+github.com/eclipse/paho.mqtt.golang v1.5.1/go.mod h1:1/yJCneuyOoCOzKSsOTUc0AJfpsItBGWvYpBLimhArU=
+github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
+github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
+github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
+github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
+github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/gorilla/websocket v1.5.3 h1:saDtZ6Pbx/0u+bgYQ3q96pZgCzfhKXGPqt7kZ72aNNg=
+github.com/gorilla/websocket v1.5.3/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
+github.com/klauspost/cpuid/v2 v2.2.10 h1:tBs3QSyvjDyFTq3uoc/9xFpCuOsJQFNPiAhYdw2skhE=
+github.com/klauspost/cpuid/v2 v2.2.10/go.mod h1:hqwkgyIinND0mEev00jJYCxPNVRVXFQeu1XKlok6oO0=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/rabbitmq/amqp091-go v1.13.0 h1:L8NA1WtF76C6KA3LAoufjfLgbist/If1UQYcsOjtxXA=
+github.com/rabbitmq/amqp091-go v1.13.0/go.mod h1:Hy4jKW5kQART1u+JkDTF9YYOQUHXqMuhrgxOEeS7G4o=
+github.com/redis/go-redis/v9 v9.21.0 h1:FPBE4hhbAke+TLmcY3WkpbDffJEomdqPn3HYiqAtL9E=
+github.com/redis/go-redis/v9 v9.21.0/go.mod h1:v/M13XI1PVCDcm01VtPFOADfZtHf8YW3baQf57KlIkA=
+github.com/stretchr/testify v1.3.0 h1:TivCn/peBQ7UY8ooIcPgZFpTNSz0Q2U6UrFlUfqbe0Q=
+github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
+github.com/yuin/gopher-lua v1.1.1 h1:kYKnWBjvbNP4XLT3+bPEwAXJx262OhaHDWDVOPjL46M=
+github.com/yuin/gopher-lua v1.1.1/go.mod h1:GBR0iDaNXjAgGg9zfCvksxSRnQx76gclCIb7kdAd1Pw=
+github.com/zeebo/xxh3 v1.1.0 h1:s7DLGDK45Dyfg7++yxI0khrfwq9661w9EN78eP/UZVs=
+github.com/zeebo/xxh3 v1.1.0/go.mod h1:IisAie1LELR4xhVinxWS5+zf1lA4p0MW4T+w+W07F5s=
+go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
+go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
+go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
+go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
+go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
+go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
+go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
+go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
+go.uber.org/atomic v1.11.0 h1:ZvwS0R+56ePWxUNi+Atn9dWONBPp/AUETXlHW0DxSjE=
+go.uber.org/atomic v1.11.0/go.mod h1:LUxbIzbOniOlMKjJjyPfpl4v+PKK2cNJn91OQbhoJI0=
+go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
+go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
+golang.org/x/net v0.53.0 h1:d+qAbo5L0orcWAr0a9JweQpjXF19LMXJE8Ey7hwOdUA=
+golang.org/x/net v0.53.0/go.mod h1:JvMuJH7rrdiCfbeHoo3fCQU24Lf5JJwT9W3sJFulfgs=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
+golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/text v0.36.0 h1:JfKh3XmcRPqZPKevfXVpI1wXPTqbkE5f7JA92a55Yxg=
+golang.org/x/text v0.36.0/go.mod h1:NIdBknypM8iqVmPiuco0Dh6P5Jcdk8lJL0CUebqK164=
+gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
+gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 h1:RmoJA1ujG+/lRGNfUnOMfhCy5EipVMyvUE+KNbPbTlw=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/grpc v1.82.1 h1:NnAxzGRA0677vCa4BUkOAnO5+FfQqVl9iUXeD0IqcGE=
+google.golang.org/grpc v1.82.1/go.mod h1:yzTZ1TB1Z3SG+LIYaI+WiE8D5+PZ3ArnrSp8zF3+/ZA=
+google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
+google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=

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

@@ -0,0 +1,90 @@
+// Package config loads device-control-service settings from environment
+// variables, matching the names used in the repo-root .env.example.
+package config
+
+import (
+	"fmt"
+	"os"
+	"strconv"
+	"time"
+)
+
+type Config struct {
+	GRPCPort int
+
+	MQTTBrokerURL string
+	MQTTClientID  string
+
+	RedisAddr     string
+	RedisPassword string
+	RedisDB       int
+
+	RabbitMQURL string
+
+	HealthCheckTimeout  time.Duration
+	HealthCheckInterval time.Duration
+}
+
+func Load() (Config, error) {
+	cfg := Config{
+		MQTTBrokerURL: fmt.Sprintf("tcp://%s:%s", getEnv("MQTT_HOST", "localhost"), getEnv("MQTT_PORT", "1883")),
+		MQTTClientID:  getEnv("DEVICE_CONTROL_MQTT_CLIENT_ID", "device-control-service"),
+
+		RedisAddr:     fmt.Sprintf("%s:%s", getEnv("REDIS_HOST", "localhost"), getEnv("REDIS_PORT", "6379")),
+		RedisPassword: getEnv("REDIS_PASSWORD", ""),
+
+		RabbitMQURL: fmt.Sprintf("amqp://%s:%s@%s:%s/",
+			getEnv("RABBITMQ_USER", "guest"),
+			getEnv("RABBITMQ_PASSWORD", "guest"),
+			getEnv("RABBITMQ_HOST", "localhost"),
+			getEnv("RABBITMQ_PORT", "5672"),
+		),
+	}
+
+	var err error
+	if cfg.GRPCPort, err = getEnvInt("DEVICE_CONTROL_GRPC_PORT", 50051); err != nil {
+		return Config{}, err
+	}
+	if cfg.RedisDB, err = getEnvInt("REDIS_DB", 0); err != nil {
+		return Config{}, err
+	}
+	if cfg.HealthCheckTimeout, err = getEnvDuration("DEVICE_CONTROL_HEALTHCHECK_TIMEOUT", 60*time.Second); err != nil {
+		return Config{}, err
+	}
+	if cfg.HealthCheckInterval, err = getEnvDuration("DEVICE_CONTROL_HEALTHCHECK_INTERVAL", 15*time.Second); err != nil {
+		return Config{}, err
+	}
+
+	return cfg, nil
+}
+
+func getEnv(key, fallback string) string {
+	if v := os.Getenv(key); v != "" {
+		return v
+	}
+	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 == "" {
+		return fallback, nil
+	}
+	d, err := time.ParseDuration(v)
+	if err != nil {
+		return 0, fmt.Errorf("%s: %w", key, err)
+	}
+	return d, nil
+}

services/device-control-service/internal/healthcheck/checker.go

@@ -0,0 +1,93 @@
+// Package healthcheck periodically flips devices to offline when they stop
+// sending telemetry/acks, per the platform's Device Shadow health-check
+// design (a goroutine + graceful shutdown, deliberately simple).
+package healthcheck
+
+import (
+	"context"
+	"log/slog"
+	"time"
+)
+
+// DeviceStore is the subset of shadow.Store the checker needs, kept as an
+// interface so tests don't require a real Redis.
+type DeviceStore interface {
+	KnownDevices(ctx context.Context) ([]string, error)
+	MarkOfflineIfStale(ctx context.Context, deviceID string, timeout time.Duration) (bool, error)
+}
+
+// OnOfflineFunc is called once per device that just transitioned to offline.
+type OnOfflineFunc func(deviceID string)
+
+type Checker struct {
+	store     DeviceStore
+	timeout   time.Duration
+	interval  time.Duration
+	onOffline OnOfflineFunc
+	logger    *slog.Logger
+
+	stop chan struct{}
+	done chan struct{}
+}
+
+func New(store DeviceStore, timeout, interval time.Duration, onOffline OnOfflineFunc, logger *slog.Logger) *Checker {
+	return &Checker{
+		store:     store,
+		timeout:   timeout,
+		interval:  interval,
+		onOffline: onOffline,
+		logger:    logger,
+	}
+}
+
+// Start begins the periodic scan loop. Call once.
+func (c *Checker) Start() {
+	c.stop = make(chan struct{})
+	c.done = make(chan struct{})
+	go c.loop()
+}
+
+// Stop signals the loop to exit and waits for it to finish.
+func (c *Checker) Stop() {
+	close(c.stop)
+	<-c.done
+}
+
+func (c *Checker) loop() {
+	defer close(c.done)
+	ticker := time.NewTicker(c.interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ticker.C:
+			c.checkAll()
+		case <-c.stop:
+			return
+		}
+	}
+}
+
+func (c *Checker) checkAll() {
+	ctx, cancel := context.WithTimeout(context.Background(), c.interval)
+	defer cancel()
+
+	ids, err := c.store.KnownDevices(ctx)
+	if err != nil {
+		c.logger.Error("healthcheck: list known devices failed", "error", err)
+		return
+	}
+
+	for _, id := range ids {
+		changed, err := c.store.MarkOfflineIfStale(ctx, id, c.timeout)
+		if err != nil {
+			c.logger.Error("healthcheck: check device failed", "device_id", id, "error", err)
+			continue
+		}
+		if changed {
+			c.logger.Info("device marked offline", "device_id", id)
+			if c.onOffline != nil {
+				c.onOffline(id)
+			}
+		}
+	}
+}

services/device-control-service/internal/healthcheck/checker_test.go

@@ -0,0 +1,76 @@
+package healthcheck
+
+import (
+	"context"
+	"io"
+	"log/slog"
+	"sync"
+	"testing"
+	"time"
+)
+
+type fakeStore struct {
+	mu      sync.Mutex
+	known   []string
+	stale   map[string]bool // deviceID -> whether MarkOfflineIfStale should report a change
+	changes []string        // devices actually marked changed, in call order
+}
+
+func (f *fakeStore) KnownDevices(context.Context) ([]string, error) {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	return append([]string(nil), f.known...), nil
+}
+
+func (f *fakeStore) MarkOfflineIfStale(_ context.Context, deviceID string, _ time.Duration) (bool, error) {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	if !f.stale[deviceID] {
+		return false, nil
+	}
+	// Only report the transition once, like the real store would.
+	f.stale[deviceID] = false
+	f.changes = append(f.changes, deviceID)
+	return true, nil
+}
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+func TestChecker_FlipsStaleDevicesAndCallsOnOffline(t *testing.T) {
+	store := &fakeStore{
+		known: []string{"d1", "d2"},
+		stale: map[string]bool{"d1": true, "d2": false},
+	}
+
+	offline := make(chan string, 2)
+	c := New(store, time.Minute, 10*time.Millisecond, func(deviceID string) {
+		offline <- deviceID
+	}, discardLogger())
+	c.Start()
+	defer c.Stop()
+
+	select {
+	case id := <-offline:
+		if id != "d1" {
+			t.Fatalf("got offline callback for %q, want d1", id)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("timed out waiting for onOffline callback")
+	}
+
+	select {
+	case id := <-offline:
+		t.Fatalf("unexpected second onOffline callback for %q (d2 was never stale)", id)
+	case <-time.After(50 * time.Millisecond):
+		// expected: no further callbacks
+	}
+}
+
+func TestChecker_StopWaitsForLoopExit(t *testing.T) {
+	store := &fakeStore{known: nil}
+	c := New(store, time.Minute, time.Hour, nil, discardLogger())
+	c.Start()
+	c.Stop() // must return without hanging
+}

services/device-control-service/internal/mqttclient/client.go

@@ -0,0 +1,107 @@
+// Package mqttclient wraps paho for device-control-service's two-way MQTT
+// use: subscribing to telemetry/ack topics for liveness and reported state,
+// and publishing commands to devices.
+package mqttclient
+
+import (
+	"fmt"
+	"log/slog"
+	"sync"
+	"time"
+
+	mqtt "github.com/eclipse/paho.mqtt.golang"
+)
+
+// Handler receives one message's topic, raw payload, and arrival time.
+type Handler func(topic string, payload []byte, receivedAt time.Time)
+
+type Config struct {
+	BrokerURL string // e.g. tcp://mosquitto:1883
+	ClientID  string
+}
+
+type subscription struct {
+	topic   string
+	qos     byte
+	handler Handler
+}
+
+// Client wraps a paho client, replaying every registered subscription on
+// each (re)connect so a dropped connection resubscribes automatically.
+type Client struct {
+	client mqtt.Client
+	logger *slog.Logger
+
+	mu   sync.Mutex
+	subs []subscription
+}
+
+func Connect(cfg Config, logger *slog.Logger) (*Client, error) {
+	c := &Client{logger: logger}
+
+	opts := mqtt.NewClientOptions().
+		AddBroker(cfg.BrokerURL).
+		SetClientID(cfg.ClientID).
+		SetAutoReconnect(true).
+		SetConnectRetry(true).
+		SetOnConnectHandler(func(mc mqtt.Client) {
+			c.resubscribeAll(mc)
+		}).
+		SetConnectionLostHandler(func(_ mqtt.Client, err error) {
+			logger.Warn("mqtt connection lost", "error", err)
+		})
+
+	mqttClient := mqtt.NewClient(opts)
+	token := mqttClient.Connect()
+	token.Wait()
+	if err := token.Error(); err != nil {
+		return nil, fmt.Errorf("connect to mqtt broker %s: %w", cfg.BrokerURL, err)
+	}
+	c.client = mqttClient
+	return c, nil
+}
+
+// Subscribe registers a handler for topic. It subscribes immediately if
+// already connected, and will be replayed on every future reconnect.
+func (c *Client) Subscribe(topic string, qos byte, handler Handler) error {
+	c.mu.Lock()
+	c.subs = append(c.subs, subscription{topic: topic, qos: qos, handler: handler})
+	c.mu.Unlock()
+
+	if !c.client.IsConnectionOpen() {
+		return nil
+	}
+	return c.subscribeOne(c.client, subscription{topic: topic, qos: qos, handler: handler})
+}
+
+func (c *Client) resubscribeAll(mc mqtt.Client) {
+	c.mu.Lock()
+	subs := append([]subscription(nil), c.subs...)
+	c.mu.Unlock()
+
+	for _, s := range subs {
+		if err := c.subscribeOne(mc, s); err != nil {
+			c.logger.Error("mqtt subscribe failed", "topic", s.topic, "error", err)
+		}
+	}
+}
+
+func (c *Client) subscribeOne(mc mqtt.Client, s subscription) error {
+	token := mc.Subscribe(s.topic, s.qos, func(_ mqtt.Client, msg mqtt.Message) {
+		s.handler(msg.Topic(), msg.Payload(), time.Now())
+	})
+	token.Wait()
+	return token.Error()
+}
+
+// Publish sends payload to topic and waits for the publish to complete.
+func (c *Client) Publish(topic string, qos byte, retained bool, payload []byte) error {
+	token := c.client.Publish(topic, qos, retained, payload)
+	token.Wait()
+	return token.Error()
+}
+
+// Close disconnects, waiting up to 250ms for in-flight work to settle.
+func (c *Client) Close() {
+	c.client.Disconnect(250)
+}

services/device-control-service/internal/rabbitmq/publisher.go

@@ -0,0 +1,78 @@
+// Package rabbitmq publishes device online/offline transitions for
+// notification-service to consume. Delivery reliability matters more than
+// latency here, so messages are persistent and the queue is durable.
+package rabbitmq
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"time"
+
+	amqp "github.com/rabbitmq/amqp091-go"
+)
+
+const StatusChangedQueue = "device.status_changed"
+
+type Publisher struct {
+	conn *amqp.Connection
+	ch   *amqp.Channel
+}
+
+func Connect(url string) (*Publisher, error) {
+	conn, err := amqp.Dial(url)
+	if err != nil {
+		return nil, fmt.Errorf("dial rabbitmq: %w", err)
+	}
+
+	ch, err := conn.Channel()
+	if err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("open channel: %w", err)
+	}
+
+	if _, err := ch.QueueDeclare(StatusChangedQueue, true, false, false, false, nil); err != nil {
+		ch.Close()
+		conn.Close()
+		return nil, fmt.Errorf("declare queue %s: %w", StatusChangedQueue, err)
+	}
+
+	return &Publisher{conn: conn, ch: ch}, nil
+}
+
+func (p *Publisher) Close() error {
+	if err := p.ch.Close(); err != nil {
+		p.conn.Close()
+		return err
+	}
+	return p.conn.Close()
+}
+
+type statusChangedEvent struct {
+	DeviceID string    `json:"device_id"`
+	Status   string    `json:"status"`
+	At       time.Time `json:"at"`
+}
+
+// PublishStatusChanged emits one event when a device transitions online or offline.
+func (p *Publisher) PublishStatusChanged(ctx context.Context, deviceID, status string) error {
+	body, err := json.Marshal(statusChangedEvent{
+		DeviceID: deviceID,
+		Status:   status,
+		At:       time.Now(),
+	})
+	if err != nil {
+		return fmt.Errorf("marshal event: %w", err)
+	}
+
+	err = p.ch.PublishWithContext(ctx, "", StatusChangedQueue, false, false, amqp.Publishing{
+		ContentType:  "application/json",
+		DeliveryMode: amqp.Persistent,
+		Timestamp:    time.Now(),
+		Body:         body,
+	})
+	if err != nil {
+		return fmt.Errorf("publish status change for device %s: %w", deviceID, err)
+	}
+	return nil
+}

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

@@ -0,0 +1,86 @@
+// Package server implements the DeviceControl gRPC service: it patches the
+// device's desired state in Redis and dispatches the command over MQTT.
+// It deliberately does not wait for the device to execute the command —
+// see the Device Shadow note in the service README.
+package server
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"log/slog"
+
+	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+)
+
+// ShadowPatcher is the subset of shadow.Store the server needs.
+type ShadowPatcher interface {
+	PatchDesiredState(ctx context.Context, deviceID string, patch map[string]any) error
+}
+
+// CommandPublisher is the subset of mqttclient.Client the server needs.
+type CommandPublisher interface {
+	Publish(topic string, qos byte, retained bool, payload []byte) error
+}
+
+type Server struct {
+	devicecontrol.UnimplementedDeviceControlServer
+
+	shadow ShadowPatcher
+	mqtt   CommandPublisher
+	logger *slog.Logger
+}
+
+func New(shadow ShadowPatcher, mqtt CommandPublisher, logger *slog.Logger) *Server {
+	return &Server{shadow: shadow, mqtt: mqtt, logger: logger}
+}
+
+func (s *Server) TurnOn(ctx context.Context, req *devicecontrol.TurnOnRequest) (*devicecontrol.CommandResult, error) {
+	return s.dispatch(ctx, 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(),
+		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(),
+		map[string]any{"level": req.GetLevel()},
+		map[string]any{"action": "set_level", "level": req.GetLevel()},
+	)
+}
+
+// dispatch is shared by all three RPCs: patch desired state first (so the UI
+// 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) {
+	if deviceID == "" {
+		return &devicecontrol.CommandResult{Success: false, Error: "device_id is required"}, nil
+	}
+
+	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
+	}
+
+	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
+	}
+
+	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
+	}
+
+	return &devicecontrol.CommandResult{Success: true}, nil
+}

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

@@ -0,0 +1,124 @@
+package server
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"io"
+	"log/slog"
+	"testing"
+
+	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+)
+
+type fakeShadow struct {
+	err     error
+	patches []map[string]any
+}
+
+func (f *fakeShadow) PatchDesiredState(_ context.Context, _ string, patch map[string]any) error {
+	if f.err != nil {
+		return f.err
+	}
+	f.patches = append(f.patches, patch)
+	return nil
+}
+
+type fakeMQTT struct {
+	err     error
+	topic   string
+	payload []byte
+}
+
+func (f *fakeMQTT) Publish(topic string, _ byte, _ bool, payload []byte) error {
+	if f.err != nil {
+		return f.err
+	}
+	f.topic = topic
+	f.payload = payload
+	return nil
+}
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+func TestTurnOn_Success(t *testing.T) {
+	sh := &fakeShadow{}
+	mq := &fakeMQTT{}
+	s := New(sh, mq, discardLogger())
+
+	res, err := s.TurnOn(context.Background(), &devicecontrol.TurnOnRequest{DeviceId: "d1"})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !res.Success {
+		t.Fatalf("expected success, got error %q", res.Error)
+	}
+	if len(sh.patches) != 1 || sh.patches[0]["power"] != "on" {
+		t.Fatalf("got desired patches %+v, want [{power: on}]", sh.patches)
+	}
+	if mq.topic != "devices/d1/commands" {
+		t.Fatalf("got topic %q, want devices/d1/commands", mq.topic)
+	}
+	var payload map[string]any
+	if err := json.Unmarshal(mq.payload, &payload); err != nil {
+		t.Fatalf("unmarshal published payload: %v", err)
+	}
+	if payload["action"] != "turn_on" {
+		t.Fatalf("got payload %+v, want action=turn_on", payload)
+	}
+}
+
+func TestSetLevel_Success(t *testing.T) {
+	sh := &fakeShadow{}
+	mq := &fakeMQTT{}
+	s := New(sh, mq, discardLogger())
+
+	res, err := s.SetLevel(context.Background(), &devicecontrol.SetLevelRequest{DeviceId: "d1", Level: 42.5})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !res.Success {
+		t.Fatalf("expected success, got error %q", res.Error)
+	}
+	if sh.patches[0]["level"] != 42.5 {
+		t.Fatalf("got desired patch %+v, want level=42.5", sh.patches[0])
+	}
+}
+
+func TestDispatch_MissingDeviceID(t *testing.T) {
+	s := New(&fakeShadow{}, &fakeMQTT{}, discardLogger())
+
+	res, err := s.TurnOff(context.Background(), &devicecontrol.TurnOffRequest{DeviceId: ""})
+	if err != nil {
+		t.Fatalf("unexpected transport error: %v", err)
+	}
+	if res.Success {
+		t.Fatal("expected failure for missing device_id")
+	}
+}
+
+func TestDispatch_ShadowFailure(t *testing.T) {
+	s := New(&fakeShadow{err: errors.New("redis down")}, &fakeMQTT{}, discardLogger())
+
+	res, err := s.TurnOn(context.Background(), &devicecontrol.TurnOnRequest{DeviceId: "d1"})
+	if err != nil {
+		t.Fatalf("unexpected transport error: %v", err)
+	}
+	if res.Success {
+		t.Fatal("expected failure when shadow patch errors")
+	}
+}
+
+func TestDispatch_PublishFailure(t *testing.T) {
+	s := New(&fakeShadow{}, &fakeMQTT{err: errors.New("broker unreachable")}, discardLogger())
+
+	res, err := s.TurnOn(context.Background(), &devicecontrol.TurnOnRequest{DeviceId: "d1"})
+	if err != nil {
+		t.Fatalf("unexpected transport error: %v", err)
+	}
+	if res.Success {
+		t.Fatal("expected failure when mqtt publish errors")
+	}
+}

services/device-control-service/internal/shadow/store.go

@@ -0,0 +1,156 @@
+// Package shadow implements the Device Shadow pattern on top of Redis:
+// desired_state (what the user wants) vs reported_state (what the device
+// confirmed), plus status and last_seen for liveness tracking.
+package shadow
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"strconv"
+	"time"
+
+	"github.com/redis/go-redis/v9"
+)
+
+// knownDevicesKey backs a lightweight device registry: health-check needs to
+// know which device IDs to scan, and this service has no PostgreSQL access
+// (device registration lives in Laravel), so every device we hear from gets
+// added to this set.
+const knownDevicesKey = "devices:known"
+
+func desiredKey(deviceID string) string  { return "device:" + deviceID + ":desired_state" }
+func reportedKey(deviceID string) string { return "device:" + deviceID + ":reported_state" }
+func statusKey(deviceID string) string   { return "device:" + deviceID + ":status" }
+func lastSeenKey(deviceID string) string { return "device:" + deviceID + ":last_seen" }
+
+const (
+	StatusOnline  = "online"
+	StatusOffline = "offline"
+)
+
+type Store struct {
+	rdb *redis.Client
+}
+
+func New(rdb *redis.Client) *Store {
+	return &Store{rdb: rdb}
+}
+
+func Connect(ctx context.Context, addr, password string, db int) (*Store, error) {
+	rdb := redis.NewClient(&redis.Options{Addr: addr, Password: password, DB: db})
+	if err := rdb.Ping(ctx).Err(); err != nil {
+		return nil, fmt.Errorf("ping redis: %w", err)
+	}
+	return New(rdb), nil
+}
+
+func (s *Store) Close() error {
+	return s.rdb.Close()
+}
+
+// PatchDesiredState merges patch into the device's desired_state JSON object.
+func (s *Store) PatchDesiredState(ctx context.Context, deviceID string, patch map[string]any) error {
+	return s.patchJSON(ctx, desiredKey(deviceID), patch)
+}
+
+// PatchReportedState merges patch into the device's reported_state JSON object.
+func (s *Store) PatchReportedState(ctx context.Context, deviceID string, patch map[string]any) error {
+	return s.patchJSON(ctx, reportedKey(deviceID), patch)
+}
+
+func (s *Store) patchJSON(ctx context.Context, key string, patch map[string]any) error {
+	existing := map[string]any{}
+	raw, err := s.rdb.Get(ctx, key).Result()
+	switch {
+	case err == redis.Nil:
+		// no existing state yet, start from an empty object
+	case err != nil:
+		return fmt.Errorf("get %s: %w", key, err)
+	default:
+		if err := json.Unmarshal([]byte(raw), &existing); err != nil {
+			return fmt.Errorf("unmarshal existing %s: %w", key, err)
+		}
+	}
+
+	for k, v := range patch {
+		existing[k] = v
+	}
+
+	merged, err := json.Marshal(existing)
+	if err != nil {
+		return fmt.Errorf("marshal %s: %w", key, err)
+	}
+	if err := s.rdb.Set(ctx, key, merged, 0).Err(); err != nil {
+		return fmt.Errorf("set %s: %w", key, err)
+	}
+	return nil
+}
+
+// Touch registers deviceID as known, bumps last_seen to now, and — if the
+// device wasn't already marked online — flips status to online. It reports
+// whether that online transition happened, so the caller can decide whether
+// a "device back online" event is worth emitting.
+func (s *Store) Touch(ctx context.Context, deviceID string) (becameOnline bool, err error) {
+	if err := s.rdb.SAdd(ctx, knownDevicesKey, deviceID).Err(); err != nil {
+		return false, fmt.Errorf("register known device %s: %w", deviceID, err)
+	}
+	if err := s.rdb.Set(ctx, lastSeenKey(deviceID), time.Now().Unix(), 0).Err(); err != nil {
+		return false, fmt.Errorf("set last_seen for %s: %w", deviceID, err)
+	}
+
+	status, err := s.rdb.Get(ctx, statusKey(deviceID)).Result()
+	if err != nil && err != redis.Nil {
+		return false, fmt.Errorf("get status for %s: %w", deviceID, err)
+	}
+	if status == StatusOnline {
+		return false, nil
+	}
+	if err := s.rdb.Set(ctx, statusKey(deviceID), StatusOnline, 0).Err(); err != nil {
+		return false, fmt.Errorf("set status online for %s: %w", deviceID, err)
+	}
+	return true, nil
+}
+
+// MarkOfflineIfStale flips deviceID to offline if it is currently online and
+// its last_seen is older than timeout. It reports whether a change was made.
+func (s *Store) MarkOfflineIfStale(ctx context.Context, deviceID string, timeout time.Duration) (changed bool, err error) {
+	status, err := s.rdb.Get(ctx, statusKey(deviceID)).Result()
+	if err == redis.Nil || status != StatusOnline {
+		return false, nil
+	}
+	if err != nil {
+		return false, fmt.Errorf("get status for %s: %w", deviceID, err)
+	}
+
+	lastSeenRaw, err := s.rdb.Get(ctx, lastSeenKey(deviceID)).Result()
+	if err != nil && err != redis.Nil {
+		return false, fmt.Errorf("get last_seen for %s: %w", deviceID, err)
+	}
+
+	var lastSeen time.Time
+	if lastSeenRaw != "" {
+		sec, err := strconv.ParseInt(lastSeenRaw, 10, 64)
+		if err != nil {
+			return false, fmt.Errorf("parse last_seen for %s: %w", deviceID, err)
+		}
+		lastSeen = time.Unix(sec, 0)
+	}
+
+	if time.Since(lastSeen) <= timeout {
+		return false, nil
+	}
+	if err := s.rdb.Set(ctx, statusKey(deviceID), StatusOffline, 0).Err(); err != nil {
+		return false, fmt.Errorf("set status offline for %s: %w", deviceID, err)
+	}
+	return true, nil
+}
+
+// KnownDevices lists every device ID this instance has ever heard from.
+func (s *Store) KnownDevices(ctx context.Context) ([]string, error) {
+	ids, err := s.rdb.SMembers(ctx, knownDevicesKey).Result()
+	if err != nil {
+		return nil, fmt.Errorf("list known devices: %w", err)
+	}
+	return ids, nil
+}

services/device-control-service/internal/shadow/store_test.go

@@ -0,0 +1,119 @@
+package shadow
+
+import (
+	"context"
+	"encoding/json"
+	"testing"
+	"time"
+
+	"github.com/alicebob/miniredis/v2"
+	"github.com/redis/go-redis/v9"
+)
+
+func newTestStore(t *testing.T) *Store {
+	t.Helper()
+	mr := miniredis.RunT(t)
+	rdb := redis.NewClient(&redis.Options{Addr: mr.Addr()})
+	t.Cleanup(func() { rdb.Close() })
+	return New(rdb)
+}
+
+func TestPatchDesiredState_MergesFields(t *testing.T) {
+	ctx := context.Background()
+	s := newTestStore(t)
+
+	if err := s.PatchDesiredState(ctx, "d1", map[string]any{"power": "on"}); err != nil {
+		t.Fatalf("first patch: %v", err)
+	}
+	if err := s.PatchDesiredState(ctx, "d1", map[string]any{"level": 42.0}); err != nil {
+		t.Fatalf("second patch: %v", err)
+	}
+
+	raw, err := s.rdb.Get(ctx, desiredKey("d1")).Result()
+	if err != nil {
+		t.Fatalf("get: %v", err)
+	}
+	var got map[string]any
+	if err := json.Unmarshal([]byte(raw), &got); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if got["power"] != "on" || got["level"] != 42.0 {
+		t.Fatalf("got %+v, want power=on and level=42", got)
+	}
+}
+
+func TestTouch_FirstCallReportsOnlineTransition(t *testing.T) {
+	ctx := context.Background()
+	s := newTestStore(t)
+
+	becameOnline, err := s.Touch(ctx, "d1")
+	if err != nil {
+		t.Fatalf("touch: %v", err)
+	}
+	if !becameOnline {
+		t.Fatal("expected first touch to report an online transition")
+	}
+
+	becameOnline, err = s.Touch(ctx, "d1")
+	if err != nil {
+		t.Fatalf("second touch: %v", err)
+	}
+	if becameOnline {
+		t.Fatal("expected second touch (already online) to report no transition")
+	}
+
+	ids, err := s.KnownDevices(ctx)
+	if err != nil {
+		t.Fatalf("known devices: %v", err)
+	}
+	if len(ids) != 1 || ids[0] != "d1" {
+		t.Fatalf("got known devices %+v, want [d1]", ids)
+	}
+}
+
+func TestMarkOfflineIfStale(t *testing.T) {
+	ctx := context.Background()
+	s := newTestStore(t)
+
+	if _, err := s.Touch(ctx, "d1"); err != nil {
+		t.Fatalf("touch: %v", err)
+	}
+
+	changed, err := s.MarkOfflineIfStale(ctx, "d1", time.Hour)
+	if err != nil {
+		t.Fatalf("mark offline (not stale): %v", err)
+	}
+	if changed {
+		t.Fatal("device just touched should not be considered stale")
+	}
+
+	changed, err = s.MarkOfflineIfStale(ctx, "d1", -time.Second) // any age counts as stale
+	if err != nil {
+		t.Fatalf("mark offline (stale): %v", err)
+	}
+	if !changed {
+		t.Fatal("expected status to flip to offline")
+	}
+
+	// Already offline: a second call should report no further change.
+	changed, err = s.MarkOfflineIfStale(ctx, "d1", -time.Second)
+	if err != nil {
+		t.Fatalf("mark offline (already offline): %v", err)
+	}
+	if changed {
+		t.Fatal("expected no change once already offline")
+	}
+}
+
+func TestMarkOfflineIfStale_UnknownDeviceIsNoop(t *testing.T) {
+	ctx := context.Background()
+	s := newTestStore(t)
+
+	changed, err := s.MarkOfflineIfStale(ctx, "ghost", time.Second)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if changed {
+		t.Fatal("unknown device should never be reported as changed")
+	}
+}

services/esp32-emulator/README.md

@@ -1,11 +1,11 @@
 # esp32-emulator (Go)
 
-Status: not implemented yet.
+Статус: пока не реализован.
 
-Responsibility:
+Зона ответственности:
-- Stand in for real ESP32 hardware during MVP development.
+- Заменяет реальное железо ESP32 на время разработки MVP.
-- Publish plausible telemetry (temperature/humidity/etc.) to
+- Публикует правдоподобную телеметрию (температура/влажность/и т.д.) в
-  `devices/{device_id}/telemetry` over MQTT on an interval.
+  `devices/{device_id}/telemetry` по MQTT с заданным интервалом.
-- Subscribe to `devices/{device_id}/commands` and reply on
+- Подписывается на `devices/{device_id}/commands` и отвечает в
-  `devices/{device_id}/ack`, so device-control-service has something real
+  `devices/{device_id}/ack`, чтобы device-control-service было с кем
-  to talk to.
+  по-настоящему взаимодействовать.

services/health-check-service/README.md

@@ -1,11 +1,12 @@
 # health-check-service (Go)
 
-Status: not implemented yet. May start as a goroutine inside
+Статус: пока не реализован. Может начаться как горутина внутри
-device-control-service and get split out into its own container later —
+device-control-service и позже выделиться в отдельный контейнер — поэтому
-kept as a separate directory from day one so the split is a non-event.
+с первого дня держим отдельную директорию, чтобы это разделение прошло
+безболезненно.
 
-Responsibility:
+Зона ответственности:
-- Periodically (ticker) check `last_seen` for every device in Redis.
+- Периодически (по тикеру) проверяет `last_seen` каждого устройства в Redis.
-- When a device exceeds the offline timeout, flip its `status` to `offline`
+- Когда устройство превышает таймаут офлайна, переводит его `status` в
-  and publish an event for notification-service.
+  `offline` и публикует событие для notification-service.
-- Good place to demonstrate goroutines + graceful shutdown in Go.
+- Хороший повод продемонстрировать горутины + graceful shutdown в Go.

services/ingest-service/README.md

@@ -1,26 +1,26 @@
 # ingest-service (Go)
 
-Status: implemented (MVP).
+Статус: реализован (MVP).
 
-Responsibility:
+Зона ответственности:
-- Subscribe to MQTT topics `devices/{device_id}/telemetry`.
+- Подписка на MQTT-топики `devices/{device_id}/telemetry`.
-- Validate/parse payload (device_id, sensor_type, value, timestamp).
+- Валидация/парсинг payload (device_id, sensor_type, value, timestamp).
-- Batch-write raw readings into ClickHouse `telemetry` table.
+- Батч-запись сырых показаний в таблицу ClickHouse `telemetry`.
-- Publish a "new reading" event to RabbitMQ for rule-engine-service.
+- Публикация события «новое показание» в RabbitMQ для rule-engine-service.
 
-Not this service's job: interpreting what a sensor reading *means* (thresholds,
+Не входит в зону ответственности: интерпретация того, что *означает*
-actions) — that belongs to rule-engine-service. This service only ingests and
+показание датчика (пороги, действия) — это задача rule-engine-service. Этот
-stores.
+сервис только принимает и сохраняет данные.
 
-Note on `zone_id`: the telemetry payload carries `zone_id` directly (the
+Про `zone_id`: payload телеметрии несёт `zone_id` напрямую (публикующее
-publishing device/emulator is configured with its own zone), rather than
+устройство/эмулятор само знает свою зону), а не ingest-service ищет его в
-ingest-service looking it up from PostgreSQL. This keeps ingest-service's
+PostgreSQL. Это удерживает зависимости ingest-service в рамках
-dependencies limited to MQTT/ClickHouse/RabbitMQ, matching the architecture
+MQTT/ClickHouse/RabbitMQ, как и на диаграмме архитектуры в корневом README.
-diagram in the root README. A real fleet where devices don't know their own
+Для реального парка устройств, которые не будут знать свою зону,
-zone would need a device-registry lookup (cached, like rule-engine caches
+понадобится lookup через реестр устройств (с кэшированием, как rule-engine
-rules) — that's a natural follow-up, not implemented yet.
+кэширует правила) — это естественное развитие, пока не реализовано.
 
-## Running
+## Запуск
 
 ```bash
 cd services/ingest-service
@@ -28,7 +28,8 @@ go test ./...
 go build ./cmd/ingest-service
 ```
 
-Configuration is env-var driven (see the ingest-service section of the
+Конфигурация читается из переменных окружения (см. секцию ingest-service в
-repo-root `.env.example`). Via `docker compose up ingest-service` it talks to
+корневом `.env.example`). Через `docker compose up ingest-service` сервис
-the `mosquitto`/`clickhouse`/`rabbitmq` containers directly; connect from a
+обращается напрямую к контейнерам `mosquitto`/`clickhouse`/`rabbitmq`; для
-local `go run` by exporting `MQTT_HOST=localhost`, `CLICKHOUSE_HOST=localhost`, etc.
+локального запуска через `go run` выставь `MQTT_HOST=localhost`,
+`CLICKHOUSE_HOST=localhost` и т.д.

services/rule-engine-service/Dockerfile

@@ -0,0 +1,15 @@
+# Build context is the repo root (not this directory) — rule-engine-service
+# depends on the sibling proto/ Go module via a relative `replace` directive.
+FROM golang:1.25-alpine AS build
+WORKDIR /src
+COPY proto/ ./proto/
+COPY services/rule-engine-service/ ./services/rule-engine-service/
+WORKDIR /src/services/rule-engine-service
+RUN go mod download
+RUN CGO_ENABLED=0 go build -o /out/rule-engine-service ./cmd/rule-engine-service
+
+FROM alpine:3.20
+RUN adduser -D -u 10001 app
+COPY --from=build /out/rule-engine-service /usr/local/bin/rule-engine-service
+USER app
+ENTRYPOINT ["/usr/local/bin/rule-engine-service"]

services/rule-engine-service/README.md

@@ -1,18 +1,52 @@
 # rule-engine-service (Go)
 
-Status: not implemented yet.
+Статус: реализован (MVP).
 
-Responsibility:
+Зона ответственности:
-- Consume "new reading" events from RabbitMQ (async — delivery reliability
+- Потребляет события «новое показание» из RabbitMQ (очередь
-  matters more than latency here).
+  `telemetry.new_reading`, публикуемая ingest-service) — асинхронно, с ручным
-- Load active `automation_rules` for the relevant zone/device from PostgreSQL,
+  ack/nack: надёжность доставки здесь важнее задержки.
-  caching in memory/Redis to avoid a DB round-trip per event.
+- Загружает активные `automation_rules` из PostgreSQL в кэш в памяти
-- Evaluate rule conditions generically: `sensor_type X operator value` against
+  (обновляется по тикеру), чтобы не ходить в БД на каждое событие.
-  the reading — no hardcoded sensor/device names, everything comes from the
+- Оценивает условия правил обобщённо: `sensor_type X operator value` против
-  `automation_rules` row.
+  показания — никаких зашитых названий датчиков/устройств, всё берётся из
-- On match: call device-control-service over gRPC (needs a fast pass/fail
+  строки `automation_rules`.
-  result) and publish an event to notification-service over RabbitMQ (not
+- При срабатывании: вызывает device-control-service по gRPC (нужен быстрый
-  latency-critical).
+  результат успех/неудача) и публикует событие `automation.rule_triggered` в
+  RabbitMQ для будущего notification-service (не критично к задержке).
 
-Not this service's job: talking to MQTT or Redis directly — device state
+Не входит в зону ответственности: прямое взаимодействие с MQTT или Redis —
-changes always go through device-control-service's gRPC API.
+изменения состояния устройства всегда идут через gRPC API device-control-service.
+
+## Архитектурные решения
+
+- **automation_rules хранит внутренние ID Postgres, а не внешние device_id.**
+  `condition_source_device_id`/`target_device_id` — это FK на `devices.id`
+  (bigserial), а MQTT/Redis/gRPC везде оперируют `devices.external_id`
+  (строка). Поэтому кэш правил грузится JOIN'ом `automation_rules` с
+  `devices` (дважды — для source и target), резолвя оба ID в external_id
+  один раз при загрузке, а не на каждое событие.
+- **Retry — только на транспортных ошибках.** Если gRPC-вызов
+  device-control-service не удался физически (сеть, сервис недоступен) —
+  событие из RabbitMQ nack'ается с `requeue=true`, всё правило
+  переигрывается позже. Если устройство само отклонило команду
+  (`CommandResult.success=false`, например неизвестный device_id) — это
+  финальный исход, ack, повторов не будет. Осознанное упрощение: если из
+  нескольких правил на одно показание одно не удалось из-за сети, при
+  повторной доставке переиграются ВСЕ правила читающие это показание, включая
+  уже успешно сработавшие — дедупликация не реализована, для MVP это
+  приемлемо (команды идемпотентны: turn_on дважды — не проблема).
+- **Кэш правил — только для чтения активных правил**, никакой записи назад в
+  Postgres. CRUD правил — это зона ответственности Laravel (этап 2).
+
+## Запуск
+
+```bash
+cd services/rule-engine-service
+go test ./...
+go build ./cmd/rule-engine-service
+```
+
+Конфигурация — через переменные окружения (секция rule-engine-service в
+корневом `.env.example`). Как и device-control-service, использует
+`proto/device_control` через `replace` на `../../proto` в `go.mod`.

services/rule-engine-service/cmd/rule-engine-service/main.go

@@ -0,0 +1,88 @@
+// Command rule-engine-service consumes telemetry readings from RabbitMQ,
+// evaluates them against PostgreSQL-backed automation rules, and dispatches
+// matching actions to device-control-service over gRPC.
+package main
+
+import (
+	"context"
+	"log/slog"
+	"os"
+	"os/signal"
+	"syscall"
+
+	"github.com/jackc/pgx/v5/pgxpool"
+
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/config"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/devicecontrolclient"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/engine"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rabbitmq"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rulecache"
+)
+
+func main() {
+	logger := slog.New(slog.NewJSONHandler(os.Stdout, nil))
+
+	if err := run(logger); err != nil {
+		logger.Error("rule-engine-service exited with error", "error", err)
+		os.Exit(1)
+	}
+}
+
+func run(logger *slog.Logger) error {
+	cfg, err := config.Load()
+	if err != nil {
+		return err
+	}
+
+	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
+	defer stop()
+
+	pgPool, err := pgxpool.New(ctx, cfg.PostgresDSN)
+	if err != nil {
+		return err
+	}
+	defer pgPool.Close()
+
+	dcClient, err := devicecontrolclient.Connect(cfg.DeviceControlGRPCTarget)
+	if err != nil {
+		return err
+	}
+	defer dcClient.Close()
+
+	mq, err := rabbitmq.Connect(cfg.RabbitMQURL)
+	if err != nil {
+		return err
+	}
+	defer mq.Close()
+
+	cache := rulecache.New(rulecache.NewPostgresFetcher(pgPool), cfg.RuleCacheRefreshInterval, logger)
+	if err := cache.Start(ctx); err != nil {
+		return err
+	}
+	defer cache.Stop()
+
+	eng := engine.New(cache, dcClient, mq, logger)
+
+	logger.Info("rule-engine-service started")
+
+	err = mq.ConsumeReadings(ctx, func(ctx context.Context, event rabbitmq.ReadingEvent) rabbitmq.HandleResult {
+		reading := engine.Reading{
+			DeviceID:   event.DeviceID,
+			ZoneID:     event.ZoneID,
+			SensorType: event.SensorType,
+			Value:      event.Value,
+			RecordedAt: event.RecordedAt,
+		}
+		if err := eng.HandleReading(ctx, reading); err != nil {
+			logger.Error("handling reading failed, will retry", "device_id", event.DeviceID, "error", err)
+			return rabbitmq.NackRequeue
+		}
+		return rabbitmq.Ack
+	})
+	if err != nil {
+		return err
+	}
+
+	logger.Info("shutting down")
+	return nil
+}

services/rule-engine-service/go.mod

@@ -0,0 +1,24 @@
+module git.cactoz.su/cacto/home_automatization/services/rule-engine-service
+
+go 1.25.0
+
+require (
+	git.cactoz.su/cacto/home_automatization/proto v0.0.0-00010101000000-000000000000
+	github.com/jackc/pgx/v5 v5.10.0
+	github.com/rabbitmq/amqp091-go v1.13.0
+	google.golang.org/grpc v1.82.1
+)
+
+require (
+	github.com/jackc/pgpassfile v1.0.0 // indirect
+	github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
+	github.com/jackc/puddle/v2 v2.2.2 // indirect
+	golang.org/x/net v0.53.0 // indirect
+	golang.org/x/sync v0.20.0 // indirect
+	golang.org/x/sys v0.43.0 // indirect
+	golang.org/x/text v0.36.0 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 // indirect
+	google.golang.org/protobuf v1.36.11 // indirect
+)
+
+replace git.cactoz.su/cacto/home_automatization/proto => ../../proto

services/rule-engine-service/go.sum

@@ -0,0 +1,66 @@
+github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
+github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
+github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
+github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
+github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
+github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
+github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
+github.com/jackc/pgx/v5 v5.10.0 h1:VhSvgU2jSli8o3AqIEOTJr7rZwAEUVo4E4XhR94Zfr0=
+github.com/jackc/pgx/v5 v5.10.0/go.mod h1:mal1tBGAFfLHvZzaYh77YS/eC6IX9OWbRV1QIIM0Jn4=
+github.com/jackc/puddle/v2 v2.2.2 h1:PR8nw+E/1w0GLuRFSmiioY6UooMp6KJv0/61nB7icHo=
+github.com/jackc/puddle/v2 v2.2.2/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/rabbitmq/amqp091-go v1.13.0 h1:L8NA1WtF76C6KA3LAoufjfLgbist/If1UQYcsOjtxXA=
+github.com/rabbitmq/amqp091-go v1.13.0/go.mod h1:Hy4jKW5kQART1u+JkDTF9YYOQUHXqMuhrgxOEeS7G4o=
+github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
+github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
+github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
+go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
+go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
+go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
+go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
+go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
+go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
+go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
+go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
+go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
+golang.org/x/net v0.53.0 h1:d+qAbo5L0orcWAr0a9JweQpjXF19LMXJE8Ey7hwOdUA=
+golang.org/x/net v0.53.0/go.mod h1:JvMuJH7rrdiCfbeHoo3fCQU24Lf5JJwT9W3sJFulfgs=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
+golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/text v0.36.0 h1:JfKh3XmcRPqZPKevfXVpI1wXPTqbkE5f7JA92a55Yxg=
+golang.org/x/text v0.36.0/go.mod h1:NIdBknypM8iqVmPiuco0Dh6P5Jcdk8lJL0CUebqK164=
+gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
+gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478 h1:RmoJA1ujG+/lRGNfUnOMfhCy5EipVMyvUE+KNbPbTlw=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260414002931-afd174a4e478/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/grpc v1.82.1 h1:NnAxzGRA0677vCa4BUkOAnO5+FfQqVl9iUXeD0IqcGE=
+google.golang.org/grpc v1.82.1/go.mod h1:yzTZ1TB1Z3SG+LIYaI+WiE8D5+PZ3ArnrSp8zF3+/ZA=
+google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
+google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
+gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
+gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

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

@@ -0,0 +1,70 @@
+// Package config loads rule-engine-service settings from environment
+// variables, matching the names used in the repo-root .env.example.
+package config
+
+import (
+	"fmt"
+	"os"
+	"time"
+)
+
+type Config struct {
+	PostgresDSN string
+
+	DeviceControlGRPCTarget string
+
+	RabbitMQURL string
+
+	RuleCacheRefreshInterval time.Duration
+}
+
+func Load() (Config, error) {
+	cfg := Config{
+		PostgresDSN: fmt.Sprintf("postgres://%s:%s@%s:%s/%s",
+			getEnv("POSTGRES_USER", "home_automation"),
+			getEnv("POSTGRES_PASSWORD", ""),
+			getEnv("POSTGRES_HOST", "localhost"),
+			getEnv("POSTGRES_PORT", "5432"),
+			getEnv("POSTGRES_DB", "home_automation"),
+		),
+
+		DeviceControlGRPCTarget: fmt.Sprintf("%s:%s",
+			getEnv("DEVICE_CONTROL_HOST", "localhost"),
+			getEnv("DEVICE_CONTROL_GRPC_PORT", "50051"),
+		),
+
+		RabbitMQURL: fmt.Sprintf("amqp://%s:%s@%s:%s/",
+			getEnv("RABBITMQ_USER", "guest"),
+			getEnv("RABBITMQ_PASSWORD", "guest"),
+			getEnv("RABBITMQ_HOST", "localhost"),
+			getEnv("RABBITMQ_PORT", "5672"),
+		),
+	}
+
+	interval, err := getEnvDuration("RULE_ENGINE_CACHE_REFRESH_INTERVAL", 15*time.Second)
+	if err != nil {
+		return Config{}, err
+	}
+	cfg.RuleCacheRefreshInterval = interval
+
+	return cfg, nil
+}
+
+func getEnv(key, fallback string) string {
+	if v := os.Getenv(key); v != "" {
+		return v
+	}
+	return fallback
+}
+
+func getEnvDuration(key string, fallback time.Duration) (time.Duration, error) {
+	v := os.Getenv(key)
+	if v == "" {
+		return fallback, nil
+	}
+	d, err := time.ParseDuration(v)
+	if err != nil {
+		return 0, fmt.Errorf("%s: %w", key, err)
+	}
+	return d, nil
+}

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

@@ -0,0 +1,95 @@
+// Package devicecontrolclient wraps the generated DeviceControl gRPC client,
+// translating an automation_rules row's generic action_type/action_params
+// into the specific TurnOn/TurnOff/SetLevel RPC.
+package devicecontrolclient
+
+import (
+	"context"
+	"fmt"
+
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/credentials/insecure"
+
+	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+)
+
+type Client struct {
+	conn   *grpc.ClientConn
+	client devicecontrol.DeviceControlClient
+}
+
+func Connect(target string) (*Client, error) {
+	conn, err := grpc.NewClient(target, grpc.WithTransportCredentials(insecure.NewCredentials()))
+	if err != nil {
+		return nil, fmt.Errorf("connect to device-control-service at %s: %w", target, err)
+	}
+	return &Client{conn: conn, client: devicecontrol.NewDeviceControlClient(conn)}, nil
+}
+
+func (c *Client) Close() error {
+	return c.conn.Close()
+}
+
+// newWithClient builds a Client around an already-constructed gRPC client,
+// bypassing Connect. Used by tests to inject a fake devicecontrol.DeviceControlClient.
+func newWithClient(client devicecontrol.DeviceControlClient) *Client {
+	return &Client{client: client}
+}
+
+// Result mirrors devicecontrol.CommandResult: Success/Error come from the
+// device-control-service's business logic, not a transport-level failure.
+type Result struct {
+	Success bool
+	Error   string
+}
+
+// Dispatch issues actionType against deviceID, using params for actions that
+// need them (currently just set_level's "level"). The returned error is only
+// set for transport/protocol failures; a rejected command comes back as
+// Result{Success: false, Error: "..."}.
+func (c *Client) Dispatch(ctx context.Context, deviceID, actionType string, params map[string]any) (Result, error) {
+	switch actionType {
+	case "turn_on":
+		res, err := c.client.TurnOn(ctx, &devicecontrol.TurnOnRequest{DeviceId: deviceID})
+		if err != nil {
+			return Result{}, fmt.Errorf("turn_on rpc: %w", err)
+		}
+		return Result{Success: res.GetSuccess(), Error: res.GetError()}, nil
+
+	case "turn_off":
+		res, err := c.client.TurnOff(ctx, &devicecontrol.TurnOffRequest{DeviceId: deviceID})
+		if err != nil {
+			return Result{}, fmt.Errorf("turn_off rpc: %w", err)
+		}
+		return Result{Success: res.GetSuccess(), Error: res.GetError()}, nil
+
+	case "set_level":
+		level, ok := numericParam(params, "level")
+		if !ok {
+			return Result{}, fmt.Errorf("set_level requires a numeric %q param, got %+v", "level", params)
+		}
+		res, err := c.client.SetLevel(ctx, &devicecontrol.SetLevelRequest{DeviceId: deviceID, Level: level})
+		if err != nil {
+			return Result{}, fmt.Errorf("set_level rpc: %w", err)
+		}
+		return Result{Success: res.GetSuccess(), Error: res.GetError()}, nil
+
+	default:
+		return Result{}, fmt.Errorf("unknown action_type %q", actionType)
+	}
+}
+
+func numericParam(params map[string]any, key string) (float64, bool) {
+	v, ok := params[key]
+	if !ok {
+		return 0, false
+	}
+	switch n := v.(type) {
+	case float64:
+		return n, true
+	case int:
+		return float64(n), true
+	default:
+		return 0, false
+	}
+}

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

@@ -0,0 +1,94 @@
+package devicecontrolclient
+
+import (
+	"context"
+	"testing"
+
+	"google.golang.org/grpc"
+
+	devicecontrol "git.cactoz.su/cacto/home_automatization/proto/device_control"
+)
+
+type fakeGRPCClient struct {
+	lastTurnOn   *devicecontrol.TurnOnRequest
+	lastTurnOff  *devicecontrol.TurnOffRequest
+	lastSetLevel *devicecontrol.SetLevelRequest
+	result       *devicecontrol.CommandResult
+	err          error
+}
+
+func (f *fakeGRPCClient) TurnOn(_ context.Context, in *devicecontrol.TurnOnRequest, _ ...grpc.CallOption) (*devicecontrol.CommandResult, error) {
+	f.lastTurnOn = in
+	return f.result, f.err
+}
+
+func (f *fakeGRPCClient) TurnOff(_ context.Context, in *devicecontrol.TurnOffRequest, _ ...grpc.CallOption) (*devicecontrol.CommandResult, error) {
+	f.lastTurnOff = in
+	return f.result, f.err
+}
+
+func (f *fakeGRPCClient) SetLevel(_ context.Context, in *devicecontrol.SetLevelRequest, _ ...grpc.CallOption) (*devicecontrol.CommandResult, error) {
+	f.lastSetLevel = in
+	return f.result, f.err
+}
+
+func TestDispatch_TurnOn(t *testing.T) {
+	fake := &fakeGRPCClient{result: &devicecontrol.CommandResult{Success: true}}
+	c := newWithClient(fake)
+
+	res, err := c.Dispatch(context.Background(), "fan-1", "turn_on", nil)
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !res.Success {
+		t.Fatalf("got %+v, want success", res)
+	}
+	if fake.lastTurnOn == nil || fake.lastTurnOn.DeviceId != "fan-1" {
+		t.Fatalf("got TurnOn request %+v, want device_id=fan-1", fake.lastTurnOn)
+	}
+}
+
+func TestDispatch_SetLevel(t *testing.T) {
+	fake := &fakeGRPCClient{result: &devicecontrol.CommandResult{Success: true}}
+	c := newWithClient(fake)
+
+	res, err := c.Dispatch(context.Background(), "light-1", "set_level", map[string]any{"level": 42.5})
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if !res.Success {
+		t.Fatalf("got %+v, want success", res)
+	}
+	if fake.lastSetLevel == nil || fake.lastSetLevel.Level != 42.5 {
+		t.Fatalf("got SetLevel request %+v, want level=42.5", fake.lastSetLevel)
+	}
+}
+
+func TestDispatch_SetLevel_MissingParam(t *testing.T) {
+	c := newWithClient(&fakeGRPCClient{})
+
+	if _, err := c.Dispatch(context.Background(), "light-1", "set_level", nil); err == nil {
+		t.Fatal("expected error for missing level param")
+	}
+}
+
+func TestDispatch_UnknownActionType(t *testing.T) {
+	c := newWithClient(&fakeGRPCClient{})
+
+	if _, err := c.Dispatch(context.Background(), "d1", "dance", nil); err == nil {
+		t.Fatal("expected error for unknown action_type")
+	}
+}
+
+func TestDispatch_BusinessFailurePassesThrough(t *testing.T) {
+	fake := &fakeGRPCClient{result: &devicecontrol.CommandResult{Success: false, Error: "device offline"}}
+	c := newWithClient(fake)
+
+	res, err := c.Dispatch(context.Background(), "fan-1", "turn_off", nil)
+	if err != nil {
+		t.Fatalf("unexpected transport error: %v", err)
+	}
+	if res.Success || res.Error != "device offline" {
+		t.Fatalf("got %+v, want business failure passed through", res)
+	}
+}

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

@@ -0,0 +1,96 @@
+// Package engine orchestrates the core rule-engine loop: given one sensor
+// reading, find the rules that watch it, evaluate their conditions, and
+// dispatch the matching actions. It knows nothing about RabbitMQ, gRPC, or
+// PostgreSQL directly — those are injected as narrow interfaces so this
+// package is testable without any of them running.
+package engine
+
+import (
+	"context"
+	"log/slog"
+	"time"
+
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/devicecontrolclient"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
+)
+
+// Reading is the generic sensor reading the engine reacts to — no
+// growbox/device-type-specific fields, matching the platform's abstractions.
+type Reading struct {
+	DeviceID   string
+	ZoneID     string
+	SensorType string
+	Value      float64
+	RecordedAt time.Time
+}
+
+// RuleSource is the subset of rulecache.Cache the engine needs.
+type RuleSource interface {
+	MatchingRules(deviceID, sensorType string) []rules.Rule
+}
+
+// Dispatcher is the subset of devicecontrolclient.Client the engine needs.
+type Dispatcher interface {
+	Dispatch(ctx context.Context, deviceID, actionType string, params map[string]any) (devicecontrolclient.Result, error)
+}
+
+// TriggerPublisher is the subset of rabbitmq.Client the engine needs.
+type TriggerPublisher interface {
+	PublishRuleTriggered(ctx context.Context, ruleID, zoneID int64, deviceID, actionType string, success bool, errMsg string) error
+}
+
+type Engine struct {
+	rules      RuleSource
+	dispatcher Dispatcher
+	publisher  TriggerPublisher
+	logger     *slog.Logger
+}
+
+func New(ruleSource RuleSource, dispatcher Dispatcher, publisher TriggerPublisher, logger *slog.Logger) *Engine {
+	return &Engine{rules: ruleSource, dispatcher: dispatcher, publisher: publisher, logger: logger}
+}
+
+// HandleReading evaluates every rule watching (r.DeviceID, r.SensorType) and
+// dispatches the ones whose condition matches. It returns a non-nil error
+// only when a dispatch failed at the transport level (e.g.
+// device-control-service unreachable) — that's the one case worth retrying
+// 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 {
+	matched := e.rules.MatchingRules(r.DeviceID, r.SensorType)
+
+	var firstTransportErr error
+	for _, rule := range matched {
+		matchesCondition, err := rules.Evaluate(rule.ConditionOperator, r.Value, rule.ConditionValue)
+		if err != nil {
+			e.logger.Error("skipping rule with invalid condition", "rule_id", rule.ID, "error", err)
+			continue
+		}
+		if !matchesCondition {
+			continue
+		}
+
+		result, dispatchErr := e.dispatcher.Dispatch(ctx, rule.TargetDeviceID, rule.ActionType, rule.ActionParams)
+
+		success := dispatchErr == nil && result.Success
+		errMsg := result.Error
+		switch {
+		case dispatchErr != nil:
+			errMsg = dispatchErr.Error()
+			e.logger.Error("dispatch action failed", "rule_id", rule.ID, "target_device_id", rule.TargetDeviceID, "error", dispatchErr)
+			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)
+		default:
+			e.logger.Info("rule triggered", "rule_id", rule.ID, "target_device_id", rule.TargetDeviceID, "action_type", rule.ActionType)
+		}
+
+		if pubErr := e.publisher.PublishRuleTriggered(ctx, rule.ID, rule.ZoneID, rule.TargetDeviceID, rule.ActionType, success, errMsg); pubErr != nil {
+			e.logger.Error("publish rule_triggered event failed", "rule_id", rule.ID, "error", pubErr)
+		}
+	}
+
+	return firstTransportErr
+}

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

@@ -0,0 +1,156 @@
+package engine
+
+import (
+	"context"
+	"errors"
+	"io"
+	"log/slog"
+	"testing"
+
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/devicecontrolclient"
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
+)
+
+type fakeRuleSource struct {
+	rules []rules.Rule
+}
+
+func (f *fakeRuleSource) MatchingRules(_, _ string) []rules.Rule {
+	return f.rules
+}
+
+type dispatchCall struct {
+	deviceID   string
+	actionType string
+	params     map[string]any
+}
+
+type fakeDispatcher struct {
+	calls  []dispatchCall
+	result devicecontrolclient.Result
+	err    error
+}
+
+func (f *fakeDispatcher) Dispatch(_ context.Context, deviceID, actionType string, params map[string]any) (devicecontrolclient.Result, error) {
+	f.calls = append(f.calls, dispatchCall{deviceID: deviceID, actionType: actionType, params: params})
+	return f.result, f.err
+}
+
+type triggeredEvent struct {
+	ruleID     int64
+	deviceID   string
+	actionType string
+	success    bool
+	errMsg     string
+}
+
+type fakePublisher struct {
+	events []triggeredEvent
+}
+
+func (f *fakePublisher) PublishRuleTriggered(_ context.Context, ruleID, _ int64, deviceID, actionType string, success bool, errMsg string) error {
+	f.events = append(f.events, triggeredEvent{ruleID: ruleID, deviceID: deviceID, actionType: actionType, success: success, errMsg: errMsg})
+	return nil
+}
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+func TestHandleReading_NoMatchingRules(t *testing.T) {
+	dispatcher := &fakeDispatcher{}
+	publisher := &fakePublisher{}
+	e := New(&fakeRuleSource{}, dispatcher, publisher, discardLogger())
+
+	if err := e.HandleReading(context.Background(), Reading{DeviceID: "sensor-1", SensorType: "temperature", Value: 30}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(dispatcher.calls) != 0 || len(publisher.events) != 0 {
+		t.Fatalf("expected no dispatch/publish, got %+v / %+v", dispatcher.calls, publisher.events)
+	}
+}
+
+func TestHandleReading_ConditionNotMet(t *testing.T) {
+	source := &fakeRuleSource{rules: []rules.Rule{
+		{ID: 1, ConditionOperator: ">", ConditionValue: 28, TargetDeviceID: "fan-1", ActionType: "turn_on"},
+	}}
+	dispatcher := &fakeDispatcher{}
+	e := New(source, dispatcher, &fakePublisher{}, discardLogger())
+
+	if err := e.HandleReading(context.Background(), Reading{Value: 20}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(dispatcher.calls) != 0 {
+		t.Fatalf("expected no dispatch when condition not met, got %+v", dispatcher.calls)
+	}
+}
+
+func TestHandleReading_ConditionMet_DispatchesAndPublishes(t *testing.T) {
+	source := &fakeRuleSource{rules: []rules.Rule{
+		{ID: 1, ZoneID: 7, ConditionOperator: ">", ConditionValue: 28, TargetDeviceID: "fan-1", ActionType: "turn_on"},
+	}}
+	dispatcher := &fakeDispatcher{result: devicecontrolclient.Result{Success: true}}
+	publisher := &fakePublisher{}
+	e := New(source, dispatcher, publisher, discardLogger())
+
+	if err := e.HandleReading(context.Background(), Reading{Value: 30}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	if len(dispatcher.calls) != 1 || dispatcher.calls[0].deviceID != "fan-1" || dispatcher.calls[0].actionType != "turn_on" {
+		t.Fatalf("got dispatch calls %+v, want one turn_on for fan-1", dispatcher.calls)
+	}
+	if len(publisher.events) != 1 || !publisher.events[0].success {
+		t.Fatalf("got publish events %+v, want one success event", publisher.events)
+	}
+}
+
+func TestHandleReading_BusinessRejection_NotRetried(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: false, Error: "device offline"}}
+	publisher := &fakePublisher{}
+	e := New(source, dispatcher, publisher, discardLogger())
+
+	err := e.HandleReading(context.Background(), Reading{Value: 30})
+	if err != nil {
+		t.Fatalf("business rejection should not be reported as a retryable error, got %v", err)
+	}
+	if len(publisher.events) != 1 || publisher.events[0].success || publisher.events[0].errMsg != "device offline" {
+		t.Fatalf("got publish events %+v, want one failed event with device offline", publisher.events)
+	}
+}
+
+func TestHandleReading_TransportError_IsRetryable(t *testing.T) {
+	source := &fakeRuleSource{rules: []rules.Rule{
+		{ID: 1, ConditionOperator: ">", ConditionValue: 28, TargetDeviceID: "fan-1", ActionType: "turn_on"},
+	}}
+	dispatcher := &fakeDispatcher{err: errors.New("connection refused")}
+	publisher := &fakePublisher{}
+	e := New(source, dispatcher, publisher, discardLogger())
+
+	err := e.HandleReading(context.Background(), Reading{Value: 30})
+	if err == nil {
+		t.Fatal("expected a retryable error when dispatch fails at the transport level")
+	}
+	if len(publisher.events) != 1 || publisher.events[0].success {
+		t.Fatalf("got publish events %+v, want one failed event recorded even on transport error", publisher.events)
+	}
+}
+
+func TestHandleReading_InvalidOperatorSkipsRuleButContinues(t *testing.T) {
+	source := &fakeRuleSource{rules: []rules.Rule{
+		{ID: 1, ConditionOperator: "~=", ConditionValue: 28, TargetDeviceID: "bad-rule"},
+		{ID: 2, ConditionOperator: ">", ConditionValue: 28, TargetDeviceID: "fan-1", ActionType: "turn_on"},
+	}}
+	dispatcher := &fakeDispatcher{result: devicecontrolclient.Result{Success: true}}
+	e := New(source, dispatcher, &fakePublisher{}, discardLogger())
+
+	if err := e.HandleReading(context.Background(), Reading{Value: 30}); err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+	if len(dispatcher.calls) != 1 || dispatcher.calls[0].deviceID != "fan-1" {
+		t.Fatalf("got dispatch calls %+v, want only the valid rule dispatched", dispatcher.calls)
+	}
+}

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

@@ -0,0 +1,62 @@
+// Package rabbitmq consumes ingest-service's "new reading" events and
+// publishes rule-trigger events for notification-service. One connection is
+// shared, with separate channels for consuming and publishing, since that's
+// the amqp best practice (a channel is not meant to be used concurrently for
+// unrelated flows).
+package rabbitmq
+
+import (
+	"fmt"
+
+	amqp "github.com/rabbitmq/amqp091-go"
+)
+
+const (
+	NewReadingQueue    = "telemetry.new_reading" // declared by ingest-service; declared here too, idempotently
+	RuleTriggeredQueue = "automation.rule_triggered"
+)
+
+type Client struct {
+	conn      *amqp.Connection
+	consumeCh *amqp.Channel
+	publishCh *amqp.Channel
+}
+
+func Connect(url string) (*Client, error) {
+	conn, err := amqp.Dial(url)
+	if err != nil {
+		return nil, fmt.Errorf("dial rabbitmq: %w", err)
+	}
+
+	consumeCh, err := conn.Channel()
+	if err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("open consume channel: %w", err)
+	}
+	// Process one reading at a time: simple, demonstrable backpressure —
+	// the next delivery isn't sent until the current one is acked/nacked.
+	if err := consumeCh.Qos(1, 0, false); err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("set qos: %w", err)
+	}
+	if _, err := consumeCh.QueueDeclare(NewReadingQueue, true, false, false, false, nil); err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("declare queue %s: %w", NewReadingQueue, err)
+	}
+
+	publishCh, err := conn.Channel()
+	if err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("open publish channel: %w", err)
+	}
+	if _, err := publishCh.QueueDeclare(RuleTriggeredQueue, true, false, false, false, nil); err != nil {
+		conn.Close()
+		return nil, fmt.Errorf("declare queue %s: %w", RuleTriggeredQueue, err)
+	}
+
+	return &Client{conn: conn, consumeCh: consumeCh, publishCh: publishCh}, nil
+}
+
+func (c *Client) Close() error {
+	return c.conn.Close()
+}

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

@@ -0,0 +1,65 @@
+package rabbitmq
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"time"
+)
+
+// ReadingEvent mirrors ingest-service's newReadingEvent wire format.
+type ReadingEvent struct {
+	DeviceID   string    `json:"device_id"`
+	ZoneID     string    `json:"zone_id"`
+	SensorType string    `json:"sensor_type"`
+	Value      float64   `json:"value"`
+	RecordedAt time.Time `json:"recorded_at"`
+}
+
+// HandleResult tells ConsumeReadings how to settle a delivery.
+type HandleResult int
+
+const (
+	Ack         HandleResult = iota // processed successfully
+	NackRequeue                     // transient failure (e.g. gRPC unreachable) — try again later
+	NackDiscard                     // permanent failure (e.g. bad data) — retrying won't help
+)
+
+// Handler processes one reading event and decides how it should be settled.
+type Handler func(ctx context.Context, event ReadingEvent) HandleResult
+
+// ConsumeReadings blocks, delivering messages to handler, until ctx is
+// cancelled or the delivery channel closes (e.g. connection lost).
+func (c *Client) ConsumeReadings(ctx context.Context, handler Handler) error {
+	deliveries, err := c.consumeCh.ConsumeWithContext(ctx, NewReadingQueue, "", false, false, false, false, nil)
+	if err != nil {
+		return fmt.Errorf("consume %s: %w", NewReadingQueue, err)
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			return nil
+		case d, ok := <-deliveries:
+			if !ok {
+				return fmt.Errorf("delivery channel for %s closed", NewReadingQueue)
+			}
+
+			var event ReadingEvent
+			if err := json.Unmarshal(d.Body, &event); err != nil {
+				// Poison message: no amount of retrying fixes malformed JSON.
+				_ = d.Nack(false, false)
+				continue
+			}
+
+			switch handler(ctx, event) {
+			case Ack:
+				_ = d.Ack(false)
+			case NackRequeue:
+				_ = d.Nack(false, true)
+			case NackDiscard:
+				_ = d.Nack(false, false)
+			}
+		}
+	}
+}

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

@@ -0,0 +1,48 @@
+package rabbitmq
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"time"
+
+	amqp "github.com/rabbitmq/amqp091-go"
+)
+
+type ruleTriggeredEvent struct {
+	RuleID     int64     `json:"rule_id"`
+	ZoneID     int64     `json:"zone_id"`
+	DeviceID   string    `json:"device_id"`
+	ActionType string    `json:"action_type"`
+	Success    bool      `json:"success"`
+	Error      string    `json:"error,omitempty"`
+	At         time.Time `json:"at"`
+}
+
+// PublishRuleTriggered emits one event per rule match+dispatch attempt, so
+// notification-service can (eventually) alert on it.
+func (c *Client) PublishRuleTriggered(ctx context.Context, ruleID, zoneID int64, deviceID, actionType string, success bool, errMsg string) error {
+	body, err := json.Marshal(ruleTriggeredEvent{
+		RuleID:     ruleID,
+		ZoneID:     zoneID,
+		DeviceID:   deviceID,
+		ActionType: actionType,
+		Success:    success,
+		Error:      errMsg,
+		At:         time.Now(),
+	})
+	if err != nil {
+		return fmt.Errorf("marshal event: %w", err)
+	}
+
+	err = c.publishCh.PublishWithContext(ctx, "", RuleTriggeredQueue, false, false, amqp.Publishing{
+		ContentType:  "application/json",
+		DeliveryMode: amqp.Persistent,
+		Timestamp:    time.Now(),
+		Body:         body,
+	})
+	if err != nil {
+		return fmt.Errorf("publish rule triggered event for rule %d: %w", ruleID, err)
+	}
+	return nil
+}

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

@@ -0,0 +1,139 @@
+// Package rulecache keeps active automation rules in memory, refreshed
+// periodically from PostgreSQL, so the hot path (one Redis/gRPC-free lookup
+// per incoming reading) never needs a database round-trip.
+package rulecache
+
+import (
+	"context"
+	"encoding/json"
+	"log/slog"
+	"sync"
+	"time"
+
+	"git.cactoz.su/cacto/home_automatization/services/rule-engine-service/internal/rules"
+)
+
+// Row is one automation_rules record already joined against devices, so
+// SourceExternalID/TargetExternalID are the external device IDs used by
+// MQTT/Redis/gRPC — not PostgreSQL's internal bigserial IDs.
+type Row struct {
+	RuleID              int64
+	ZoneID              int64
+	SourceExternalID    string
+	TargetExternalID    string
+	ConditionSensorType string
+	ConditionOperator   string
+	ConditionValue      float64
+	ActionType          string
+	ActionParams        []byte // raw JSON, decoded during index build
+}
+
+// Fetcher loads all currently active rules. The real implementation queries
+// PostgreSQL; tests supply a fake.
+type Fetcher interface {
+	FetchActiveRules(ctx context.Context) ([]Row, error)
+}
+
+type Cache struct {
+	fetcher  Fetcher
+	interval time.Duration
+	logger   *slog.Logger
+
+	mu    sync.RWMutex
+	index map[string][]rules.Rule
+
+	stop chan struct{}
+	done chan struct{}
+}
+
+func New(fetcher Fetcher, interval time.Duration, logger *slog.Logger) *Cache {
+	return &Cache{fetcher: fetcher, interval: interval, logger: logger}
+}
+
+// Start performs a synchronous initial load (so the service never runs with
+// an empty cache) and then begins the periodic refresh loop.
+func (c *Cache) Start(ctx context.Context) error {
+	if err := c.refresh(ctx); err != nil {
+		return err
+	}
+	c.stop = make(chan struct{})
+	c.done = make(chan struct{})
+	go c.loop()
+	return nil
+}
+
+func (c *Cache) Stop() {
+	close(c.stop)
+	<-c.done
+}
+
+func (c *Cache) loop() {
+	defer close(c.done)
+	ticker := time.NewTicker(c.interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ticker.C:
+			ctx, cancel := context.WithTimeout(context.Background(), c.interval)
+			if err := c.refresh(ctx); err != nil {
+				c.logger.Error("rule cache refresh failed", "error", err)
+			}
+			cancel()
+		case <-c.stop:
+			return
+		}
+	}
+}
+
+func (c *Cache) refresh(ctx context.Context) error {
+	dbRows, err := c.fetcher.FetchActiveRules(ctx)
+	if err != nil {
+		return err
+	}
+	index := buildIndex(dbRows, c.logger)
+	c.mu.Lock()
+	c.index = index
+	c.mu.Unlock()
+	return nil
+}
+
+// MatchingRules returns the active rules that watch sensorType readings from
+// deviceID. Safe for concurrent use.
+func (c *Cache) MatchingRules(deviceID, sensorType string) []rules.Rule {
+	c.mu.RLock()
+	defer c.mu.RUnlock()
+	return c.index[cacheKey(deviceID, sensorType)]
+}
+
+func cacheKey(deviceID, sensorType string) string {
+	return deviceID + "\x00" + sensorType
+}
+
+func buildIndex(dbRows []Row, logger *slog.Logger) map[string][]rules.Rule {
+	index := make(map[string][]rules.Rule)
+	for _, r := range dbRows {
+		var params map[string]any
+		if len(r.ActionParams) > 0 {
+			if err := json.Unmarshal(r.ActionParams, &params); err != nil {
+				logger.Error("skipping rule with invalid action_params", "rule_id", r.RuleID, "error", err)
+				continue
+			}
+		}
+
+		rule := rules.Rule{
+			ID:                  r.RuleID,
+			ZoneID:              r.ZoneID,
+			SourceDeviceID:      r.SourceExternalID,
+			ConditionSensorType: r.ConditionSensorType,
+			ConditionOperator:   r.ConditionOperator,
+			ConditionValue:      r.ConditionValue,
+			TargetDeviceID:      r.TargetExternalID,
+			ActionType:          r.ActionType,
+			ActionParams:        params,
+		}
+
+		key := cacheKey(r.SourceExternalID, r.ConditionSensorType)
+		index[key] = append(index[key], rule)
+	}
+	return index
+}

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

@@ -0,0 +1,110 @@
+package rulecache
+
+import (
+	"context"
+	"io"
+	"log/slog"
+	"sync"
+	"testing"
+	"time"
+)
+
+type fakeFetcher struct {
+	mu    sync.Mutex
+	rows  []Row
+	calls chan struct{}
+}
+
+func (f *fakeFetcher) setRows(rows []Row) {
+	f.mu.Lock()
+	defer f.mu.Unlock()
+	f.rows = rows
+}
+
+func (f *fakeFetcher) FetchActiveRules(context.Context) ([]Row, error) {
+	f.mu.Lock()
+	rows := append([]Row(nil), f.rows...)
+	f.mu.Unlock()
+	if f.calls != nil {
+		f.calls <- struct{}{}
+	}
+	return rows, nil
+}
+
+func discardLogger() *slog.Logger {
+	return slog.New(slog.NewTextHandler(io.Discard, nil))
+}
+
+func TestCache_StartLoadsSynchronously(t *testing.T) {
+	fetcher := &fakeFetcher{rows: []Row{
+		{RuleID: 1, SourceExternalID: "sensor-1", ConditionSensorType: "temperature", ConditionOperator: ">", ConditionValue: 28, TargetExternalID: "fan-1", ActionType: "turn_on"},
+	}}
+	c := New(fetcher, time.Hour, discardLogger())
+
+	if err := c.Start(context.Background()); err != nil {
+		t.Fatalf("start: %v", err)
+	}
+	defer c.Stop()
+
+	got := c.MatchingRules("sensor-1", "temperature")
+	if len(got) != 1 || got[0].TargetDeviceID != "fan-1" {
+		t.Fatalf("got %+v, want one rule targeting fan-1", got)
+	}
+
+	// Different sensor_type on the same device must not match.
+	if got := c.MatchingRules("sensor-1", "humidity"); len(got) != 0 {
+		t.Fatalf("got %+v, want no rules for humidity", got)
+	}
+}
+
+func TestCache_SkipsRuleWithInvalidActionParams(t *testing.T) {
+	fetcher := &fakeFetcher{rows: []Row{
+		{RuleID: 1, SourceExternalID: "sensor-1", ConditionSensorType: "temperature", ActionParams: []byte(`not json`)},
+		{RuleID: 2, SourceExternalID: "sensor-1", ConditionSensorType: "temperature", ActionParams: []byte(`{"level":50}`)},
+	}}
+	c := New(fetcher, time.Hour, discardLogger())
+	if err := c.Start(context.Background()); err != nil {
+		t.Fatalf("start: %v", err)
+	}
+	defer c.Stop()
+
+	got := c.MatchingRules("sensor-1", "temperature")
+	if len(got) != 1 || got[0].ID != 2 {
+		t.Fatalf("got %+v, want only rule 2 to survive", got)
+	}
+}
+
+func TestCache_RefreshesOnInterval(t *testing.T) {
+	calls := make(chan struct{}, 4)
+	fetcher := &fakeFetcher{calls: calls}
+	c := New(fetcher, 20*time.Millisecond, discardLogger())
+	if err := c.Start(context.Background()); err != nil {
+		t.Fatalf("start: %v", err)
+	}
+	defer c.Stop()
+
+	<-calls // consume the initial synchronous load
+
+	fetcher.setRows([]Row{
+		{RuleID: 1, SourceExternalID: "sensor-1", ConditionSensorType: "temperature", TargetExternalID: "fan-1", ActionType: "turn_on"},
+	})
+
+	select {
+	case <-calls:
+	case <-time.After(time.Second):
+		t.Fatal("timed out waiting for periodic refresh")
+	}
+
+	got := c.MatchingRules("sensor-1", "temperature")
+	if len(got) != 1 {
+		t.Fatalf("got %+v after refresh, want the newly added rule", got)
+	}
+}
+
+func TestCache_StopWaitsForLoopExit(t *testing.T) {
+	c := New(&fakeFetcher{}, time.Hour, discardLogger())
+	if err := c.Start(context.Background()); err != nil {
+		t.Fatalf("start: %v", err)
+	}
+	c.Stop() // must return without hanging
+}

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

@@ -0,0 +1,50 @@
+package rulecache
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/jackc/pgx/v5"
+	"github.com/jackc/pgx/v5/pgxpool"
+)
+
+// activeRulesQuery resolves condition_source_device_id/target_device_id (FKs
+// into devices.id) to devices.external_id, since every other part of the
+// platform (MQTT, Redis, gRPC) keys devices by their external ID, not
+// PostgreSQL's internal bigserial ID.
+const activeRulesQuery = `
+SELECT
+	r.id AS rule_id,
+	r.zone_id AS zone_id,
+	src.external_id AS source_external_id,
+	tgt.external_id AS target_external_id,
+	r.condition_sensor_type AS condition_sensor_type,
+	r.condition_operator AS condition_operator,
+	r.condition_value AS condition_value,
+	r.action_type AS action_type,
+	r.action_params AS action_params
+FROM automation_rules r
+JOIN devices src ON src.id = r.condition_source_device_id
+JOIN devices tgt ON tgt.id = r.target_device_id
+WHERE r.is_active = true
+`
+
+type PostgresFetcher struct {
+	pool *pgxpool.Pool
+}
+
+func NewPostgresFetcher(pool *pgxpool.Pool) *PostgresFetcher {
+	return &PostgresFetcher{pool: pool}
+}
+
+func (f *PostgresFetcher) FetchActiveRules(ctx context.Context) ([]Row, error) {
+	dbRows, err := f.pool.Query(ctx, activeRulesQuery)
+	if err != nil {
+		return nil, fmt.Errorf("query active rules: %w", err)
+	}
+	rows, err := pgx.CollectRows(dbRows, pgx.RowToStructByName[Row])
+	if err != nil {
+		return nil, fmt.Errorf("scan active rules: %w", err)
+	}
+	return rows, nil
+}

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

@@ -0,0 +1,41 @@
+// Package rules holds the generic automation-rule model and condition
+// evaluation logic. Nothing here knows about growboxes, lights, or pumps —
+// only sensor_type/operator/value, exactly as stored in automation_rules.
+package rules
+
+import "fmt"
+
+// Rule mirrors one row of automation_rules, joined against devices so the
+// device IDs are the external identifiers used by MQTT/Redis/gRPC (not
+// PostgreSQL's internal bigserial IDs).
+type Rule struct {
+	ID                  int64
+	ZoneID              int64
+	SourceDeviceID      string // condition_source_device_id, resolved to devices.external_id
+	ConditionSensorType string
+	ConditionOperator   string
+	ConditionValue      float64
+	TargetDeviceID      string // target_device_id, resolved to devices.external_id
+	ActionType          string
+	ActionParams        map[string]any
+}
+
+// Evaluate reports whether value satisfies the rule's condition.
+func Evaluate(operator string, value, threshold float64) (bool, error) {
+	switch operator {
+	case ">":
+		return value > threshold, nil
+	case "<":
+		return value < threshold, nil
+	case ">=":
+		return value >= threshold, nil
+	case "<=":
+		return value <= threshold, nil
+	case "=":
+		return value == threshold, nil
+	case "!=":
+		return value != threshold, nil
+	default:
+		return false, fmt.Errorf("unknown condition operator %q", operator)
+	}
+}

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

@@ -0,0 +1,39 @@
+package rules
+
+import "testing"
+
+func TestEvaluate(t *testing.T) {
+	cases := []struct {
+		operator  string
+		value     float64
+		threshold float64
+		want      bool
+	}{
+		{">", 29, 28, true},
+		{">", 27, 28, false},
+		{"<", 27, 28, true},
+		{"<", 29, 28, false},
+		{">=", 28, 28, true},
+		{"<=", 28, 28, true},
+		{"=", 28, 28, true},
+		{"=", 28.1, 28, false},
+		{"!=", 28.1, 28, true},
+		{"!=", 28, 28, false},
+	}
+
+	for _, c := range cases {
+		got, err := Evaluate(c.operator, c.value, c.threshold)
+		if err != nil {
+			t.Fatalf("operator %q: unexpected error: %v", c.operator, err)
+		}
+		if got != c.want {
+			t.Errorf("Evaluate(%q, %v, %v) = %v, want %v", c.operator, c.value, c.threshold, got, c.want)
+		}
+	}
+}
+
+func TestEvaluate_UnknownOperator(t *testing.T) {
+	if _, err := Evaluate("~=", 1, 2); err == nil {
+		t.Fatal("expected error for unknown operator")
+	}
+}