Добавлены метрики Prometheus и дашборды Grafana (этап 2.5)

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

monitoring/grafana/provisioning/dashboards/automation-devices.json

@@ -0,0 +1,83 @@
+{
+  "uid": "automation-devices",
+  "title": "Automation & Devices",
+  "tags": ["home-automation"],
+  "timezone": "browser",
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "10s",
+  "time": { "from": "now-30m", "to": "now" },
+  "panels": [
+    {
+      "id": 1,
+      "title": "Rules triggered (rate, by outcome)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 0 },
+      "targets": [
+        {
+          "expr": "sum by (outcome) (rate(rule_engine_rules_triggered_total[1m]))",
+          "legendFormat": "{{outcome}}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 2,
+      "title": "device-control-service dispatch duration (p95, by action)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 0 },
+      "fieldConfig": { "defaults": { "unit": "s" }, "overrides": [] },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.95, sum(rate(rule_engine_dispatch_duration_seconds_bucket[5m])) by (le, action_type))",
+          "legendFormat": "{{action_type}}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 3,
+      "title": "Device commands (rate, by action + outcome)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 8 },
+      "targets": [
+        {
+          "expr": "sum by (action, outcome) (rate(device_control_commands_total[1m]))",
+          "legendFormat": "{{action}} / {{outcome}}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 4,
+      "title": "Device online/offline transitions (rate)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 8 },
+      "targets": [
+        {
+          "expr": "sum by (direction) (rate(device_control_health_transitions_total[5m]))",
+          "legendFormat": "{{direction}}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 5,
+      "title": "Automation rule cache refreshes (rate, by outcome)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 16 },
+      "targets": [
+        {
+          "expr": "sum by (outcome) (rate(rule_engine_cache_refresh_total[1m]))",
+          "legendFormat": "{{outcome}}",
+          "refId": "A"
+        }
+      ]
+    }
+  ]
+}

monitoring/grafana/provisioning/dashboards/dashboards.yml

@@ -0,0 +1,11 @@
+apiVersion: 1
+
+providers:
+  - name: default
+    orgId: 1
+    folder: ""
+    type: file
+    disableDeletion: false
+    updateIntervalSeconds: 30
+    options:
+      path: /etc/grafana/provisioning/dashboards

monitoring/grafana/provisioning/dashboards/ingest-telemetry.json

@@ -0,0 +1,88 @@
+{
+  "uid": "ingest-telemetry",
+  "title": "Ingest & Telemetry",
+  "tags": ["home-automation"],
+  "timezone": "browser",
+  "schemaVersion": 39,
+  "version": 1,
+  "refresh": "10s",
+  "time": { "from": "now-30m", "to": "now" },
+  "panels": [
+    {
+      "id": 1,
+      "title": "MQTT messages received (rate)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 0 },
+      "targets": [
+        {
+          "expr": "sum(rate(ingest_mqtt_messages_received_total[1m]))",
+          "legendFormat": "received/s",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 2,
+      "title": "MQTT messages invalid / dropped (rate)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 0 },
+      "targets": [
+        {
+          "expr": "sum(rate(ingest_mqtt_messages_invalid_total[1m]))",
+          "legendFormat": "invalid/s",
+          "refId": "A"
+        },
+        {
+          "expr": "sum(rate(ingest_mqtt_messages_dropped_total[1m]))",
+          "legendFormat": "dropped (backlog full)/s",
+          "refId": "B"
+        }
+      ]
+    },
+    {
+      "id": 3,
+      "title": "ClickHouse batch flushes (rate, by outcome)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 8 },
+      "targets": [
+        {
+          "expr": "sum by (outcome) (rate(ingest_batch_flush_total[1m]))",
+          "legendFormat": "{{outcome}}",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 4,
+      "title": "ClickHouse batch flush duration (p95)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 8 },
+      "fieldConfig": { "defaults": { "unit": "s" }, "overrides": [] },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.95, sum(rate(ingest_batch_flush_duration_seconds_bucket[5m])) by (le))",
+          "legendFormat": "p95",
+          "refId": "A"
+        }
+      ]
+    },
+    {
+      "id": 5,
+      "title": "RabbitMQ publishes (rate, by outcome)",
+      "type": "timeseries",
+      "datasource": { "type": "prometheus", "uid": "Prometheus" },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 16 },
+      "targets": [
+        {
+          "expr": "sum by (outcome) (rate(ingest_rabbitmq_publish_total[1m]))",
+          "legendFormat": "{{outcome}}",
+          "refId": "A"
+        }
+      ]
+    }
+  ]
+}

monitoring/grafana/provisioning/datasources/datasource.yml

@@ -0,0 +1,9 @@
+apiVersion: 1
+
+datasources:
+  - name: Prometheus
+    type: prometheus
+    access: proxy
+    url: http://prometheus:9090
+    isDefault: true
+    editable: false

monitoring/prometheus/prometheus.yml

@@ -0,0 +1,15 @@
+global:
+  scrape_interval: 15s
+
+scrape_configs:
+  - job_name: ingest-service
+    static_configs:
+      - targets: ["ingest-service:9101"]
+
+  - job_name: device-control-service
+    static_configs:
+      - targets: ["device-control-service:8090"]
+
+  - job_name: rule-engine-service
+    static_configs:
+      - targets: ["rule-engine-service:9102"]