cacto
81040eec62
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").
89 lines
2.5 KiB
JSON
89 lines
2.5 KiB
JSON
{
|
|
"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"
|
|
}
|
|
]
|
|
}
|
|
]
|
|
}
|