SigNoz Observability Hexagonal Architecture —

SigNoz Hexagonal Observability

SigNoz Observability Hexagonal Architecture OpenTelemetry Traces Metrics Logs ClickHouse APM Dashboard Alerts Production

เนื้อหาเกี่ยวข้อง — อ่านต่อ: unemployment claims คือ

OpenTelemetry Setup

# === OpenTelemetry + SigNoz Setup ===

# Install SigNoz (Docker Compose)
# git clone https://github.com/SigNoz/signoz.git
# cd signoz/deploy
# docker compose -f docker/clickhouse-setup/docker-compose.yaml up -d
#
# Access Dashboard: http://localhost:3301
#
# Install OpenTelemetry SDK (Python)
# pip install opentelemetry-api opentelemetry-sdk
# pip install opentelemetry-exporter-otlp
# pip install opentelemetry-instrumentation-flask
# pip install opentelemetry-instrumentation-requests
# pip install opentelemetry-instrumentation-sqlalchemy

from dataclasses import dataclass

@dataclass
class OTelComponent:
    component: str
    hex_layer: str
    what_it_captures: str
    signoz_feature: str

components = [
    OTelComponent("Tracer (Spans)",
        "All Layers",
        "Request Flow ข้าม Service + Duration + Status",
        "Distributed Tracing → Trace Explorer"),
    OTelComponent("Meter (Metrics)",
        "All Layers",
        "Counter Histogram Gauge (Rate Error Duration)",
        "Metrics → Dashboard + Alerts"),
    OTelComponent("Logger (Logs)",
        "All Layers",
        "Structured Log + trace_id Correlation",
        "Logs → Log Explorer (Correlated with Traces)"),
    OTelComponent("Propagator (Context)",
        "Infrastructure (HTTP/gRPC)",
        "Trace Context ข้าม Service (W3C TraceContext)",
        "Service Map → Dependency Graph"),
    OTelComponent("Auto-instrumentation",
        "Infrastructure",
        "Flask Express Spring DB HTTP Client อัตโนมัติ",
        "Traces + Metrics โดยไม่ต้องเขียน Code"),
]

print("=== OpenTelemetry Components ===")
for c in components:
    print(f"  [{c.component}] Layer: {c.hex_layer}")
    print(f"    Captures: {c.what_it_captures}")
    print(f"    SigNoz: {c.signoz_feature}")

Hexagonal Instrumentation

# === Instrumentation per Hexagonal Layer ===

# Domain Layer (Business Logic)
# @tracer.start_as_current_span("domain.process_order")
# def process_order(order: Order) -> OrderResult:
#     span = trace.get_current_span()
#     span.set_attribute("order.id", order.id)
#     span.set_attribute("order.amount", order.total)
#     orders_counter.add(1, {"status": "processing"})
#     # Business logic...
#     return result
#
# Application Layer (Use Cases)
# @tracer.start_as_current_span("usecase.create_order")
# async def create_order(request: CreateOrderRequest) -> CreateOrderResponse:
#     span = trace.get_current_span()
#     span.set_attribute("user.id", request.user_id)
#     with usecase_duration.record({"usecase": "create_order"}):
#         order = Order.create(request)
#         result = process_order(order)
#         await notify_customer(result)
#     return response
#
# Infrastructure Layer (Adapters)
# Auto-instrumented by OpenTelemetry:
# - HTTP (Flask/FastAPI): Request Duration Status Code
# - Database (SQLAlchemy): Query Duration Statement
# - HTTP Client (requests): External Call Duration

@dataclass
class InstrumentationGuide:
    layer: str
    span_name: str
    attributes: str
    metrics: str
    logs: str

guides = [
    InstrumentationGuide("Domain",
        "domain.{service}.{method}",
        "order.id order.amount user.id product.id",
        "orders_total revenue_total conversion_rate",
        "OrderCreated PaymentProcessed InventoryUpdated"),
    InstrumentationGuide("Application",
        "usecase.{name}",
        "user.id request.type session.id",
        "usecase_duration usecase_errors auth_attempts",
        "UseCaseStarted UseCaseCompleted AuthFailed"),
    InstrumentationGuide("Infrastructure (DB)",
        "db.{operation} (Auto-instrumented)",
        "db.system db.statement db.operation",
        "db_query_duration db_connections_active",
        "SlowQuery ConnectionPoolExhausted"),
    InstrumentationGuide("Infrastructure (HTTP)",
        "http.{method} (Auto-instrumented)",
        "http.method http.url http.status_code",
        "http_request_duration http_requests_total",
        "HTTPError TimeoutError RateLimited"),
    InstrumentationGuide("Infrastructure (Queue)",
        "queue.{operation}",
        "messaging.system messaging.destination",
        "queue_messages_produced queue_lag",
        "MessagePublished MessageConsumed ConsumerError"),
]

print("=== Instrumentation Guide ===")
for g in guides:
    print(f"\n  [{g.layer}] Span: {g.span_name}")
    print(f"    Attributes: {g.attributes}")
    print(f"    Metrics: {g.metrics}")
    print(f"    Logs: {g.logs}")

Production Dashboard

# === SigNoz Production Setup ===

@dataclass
class DashboardPanel:
    panel: str
    query: str
    target: str
    alert: str

panels = [
    DashboardPanel("Request Rate (RPM)",
        "rate(http_requests_total[5m]) * 60",
        "Baseline ±20%",
        "Drop > 50% → P1 Service Down"),
    DashboardPanel("Error Rate (%)",
        "rate(http_requests_total{status=~'5..'}[5m]) / rate(http_requests_total[5m])",
        "< 1%",
        "> 1% → P2 | > 5% → P1"),
    DashboardPanel("P99 Latency (ms)",
        "histogram_quantile(0.99, rate(http_request_duration_bucket[5m]))",
        "< 500ms",
        "> 1000ms → P2 | > 3000ms → P1"),
    DashboardPanel("Business: Orders/min",
        "rate(orders_total[5m]) * 60",
        "Baseline ±20%",
        "Drop > 30% → P2 Check Conversion"),
    DashboardPanel("DB Query Duration P95",
        "histogram_quantile(0.95, rate(db_query_duration_bucket[5m]))",
        "< 100ms",
        "> 500ms → P2 Slow Query"),
    DashboardPanel("ClickHouse Storage",
        "signoz_storage_bytes",
        "< 80% Disk",
        "> 90% → P2 Increase Retention/Sampling"),
]

print("=== Dashboard Panels ===")
for p in panels:
    print(f"  [{p.panel}]")
    print(f"    Query: {p.query}")
    print(f"    Target: {p.target}")
    print(f"    Alert: {p.alert}")

เคล็ดลับ

• OpenTelemetry: ใช้ OpenTelemetry Standard ไม่ Lock-in Vendor
• Sampling: ใช้ Tail-based Sampling เก็บเฉพาะ Error/Slow Traces
• Correlation: ใส่ trace_id ใน Log ทุกบรรทัด Correlate Trace+Log
• Business Metrics: สร้าง Business KPI Dashboard (Orders Revenue)
• ClickHouse: ClickHouse เร็วมาก Query Trace/Metric มหาศาลได้

SigNoz คืออะไร

Open Source APM Observability Traces Metrics Logs OpenTelemetry ClickHouse Dashboard Alerts ทดแทน Datadog New Relic ถูกกว่า

แนะนำเพิ่มเติม — อ่านเพิ่มเติมที่ SiamCafeBook

เนื้อหาเกี่ยวข้อง — อ่านต่อ: CrewAI Multi-Agent Edge Computing — คู่มือฉบับสมบูรณ์ 2026

เนื้อหาเกี่ยวข้อง — ทำความเข้าใจ agile method คือ — ข้อมูลครบถ้วน 2026