service-mesh-observability

服务网格可观测性

Istio、Linkerd 和服务网格部署的可观测性模式完整指南。

不要在以下情况下使用此技能

• 任务与服务网格可观测性无关
• 你需要使用此范围之外的其他领域知识或工具

说明

• 明确目标、约束和所需输入。
• 应用相关最佳实践并验证结果。
• 提供可执行的步骤和验证方法。

在以下情况下使用此技能

• 为多个服务设置分布式追踪
• 实现服务网格指标和仪表盘
• 调试延迟和错误问题
• 为服务通信定义 SLO
• 可视化服务依赖关系
• 排查网格连通性问题

核心概念

1. 可观测性的三大支柱

┌─────────────────────────────────────────────────────┐
│                  Observability                       │
├─────────────────┬─────────────────┬─────────────────┤
│     Metrics     │     Traces      │      Logs       │
│                 │                 │                 │
│ • Request rate  │ • Span context  │ • Access logs   │
│ • Error rate    │ • Latency       │ • Error details │
│ • Latency P50   │ • Dependencies  │ • Debug info    │
│ • Saturation    │ • Bottlenecks   │ • Audit trail   │
└─────────────────┴─────────────────┴─────────────────┘

2. 服务网格黄金信号

信号	描述	告警阈值
延迟	请求耗时 P50、P99	P99 > 500ms
流量	每秒请求数	异常检测
错误	5xx 错误率	> 1%
饱和度	资源利用率	> 80%

模板

以下是特定版本的片段，并非完整的可部署堆栈。请固定服务网格、采集器和 Operator 的版本，导出或渲染受支持的架构，并在应用清单前进行审查。请单独提供 Services、RBAC 以及接收器／提供程序连接配置；用于 istiod 的 ServiceMonitor 本身不会抓取每个工作负载代理。阈值仅为示例，需要根据具体服务的 SLO 进行调整。

请确保追踪／tap 输出仅保留在获授权的环境中，对敏感属性进行脱敏并限制访问。匿名 Kiali 示例仅适用于隔离且可信的开发环境；在共享使用前，请配置经过身份验证的访问，并缩小命名空间／RBAC 范围；请参阅 Kiali 身份验证。

模板 1：使用 Prometheus 和 Grafana 的 Istio

# Install Prometheus
apiVersion: v1
kind: ConfigMap
metadata:
  name: prometheus
  namespace: istio-system
data:
  prometheus.yml: |
    global:
      scrape_interval: 15s
    scrape_configs:
      - job_name: 'istio-mesh'
        kubernetes_sd_configs:
          - role: endpoints
            namespaces:
              names:
                - istio-system
        relabel_configs:
          - source_labels: [__meta_kubernetes_service_name]
            action: keep
            regex: istio-telemetry
---
# ServiceMonitor for Prometheus Operator
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: istio-mesh
  namespace: istio-system
spec:
  selector:
    matchLabels:
      app: istiod
  endpoints:
    - port: http-monitoring
      interval: 15s

模板 2：关键 Istio 指标查询

# Request rate by service
sum(rate(istio_requests_total{reporter="destination"}[5m])) by (destination_service_name)

# Error rate (5xx)
sum(rate(istio_requests_total{reporter="destination", response_code=~"5.."}[5m]))
  / sum(rate(istio_requests_total{reporter="destination"}[5m])) * 100

# P99 latency
histogram_quantile(0.99,
  sum(rate(istio_request_duration_milliseconds_bucket{reporter="destination"}[5m]))
  by (le, destination_service_name))

# TCP connections
sum(istio_tcp_connections_opened_total{reporter="destination"}) by (destination_service_name)

# Request size
histogram_quantile(0.99,
  sum(rate(istio_request_bytes_bucket{reporter="destination"}[5m]))
  by (le, destination_service_name))

模板 3：Jaeger 分布式追踪

# Jaeger installation for Istio
apiVersion: install.istio.io/v1alpha1
kind: IstioOperator
spec:
  meshConfig:
    enableTracing: true
    defaultConfig:
      tracing:
        sampling: 100.0  # 100% in dev, lower in prod
        zipkin:
          address: jaeger-collector.istio-system:9411
---
# Jaeger deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: jaeger
  namespace: istio-system
spec:
  selector:
    matchLabels:
      app: jaeger
  template:
    metadata:
      labels:
        app: jaeger
    spec:
      containers:
        - name: jaeger
          image: jaegertracing/all-in-one:1.50
          ports:
            - containerPort: 5775   # UDP
            - containerPort: 6831   # Thrift
            - containerPort: 6832   # Thrift
            - containerPort: 5778   # Config
            - containerPort: 16686  # UI
            - containerPort: 14268  # HTTP
            - containerPort: 14250  # gRPC
            - containerPort: 9411   # Zipkin
          env:
            - name: COLLECTOR_ZIPKIN_HOST_PORT
              value: ":9411"

模板 4：Linkerd Viz 仪表盘

# Install Linkerd viz extension
linkerd viz install | kubectl apply -f -

# Access dashboard
linkerd viz dashboard

# CLI commands for observability
# Top requests
linkerd viz top deploy/my-app

# Per-route metrics
linkerd viz routes deploy/my-app --to deploy/backend

# Live traffic inspection
linkerd viz tap deploy/my-app --to deploy/backend

# Service edges (dependencies)
linkerd viz edges deployment -n my-namespace

模板 5：Grafana 仪表盘 JSON

{
  "dashboard": {
    "title": "Service Mesh Overview",
    "panels": [
      {
        "title": "Request Rate",
        "type": "graph",
        "targets": [
          {
            "expr": "sum(rate(istio_requests_total{reporter=\"destination\"}[5m])) by (destination_service_name)",
            "legendFormat": "{{destination_service_name}}"
          }
        ]
      },
      {
        "title": "Error Rate",
        "type": "gauge",
        "targets": [
          {
            "expr": "sum(rate(istio_requests_total{response_code=~\"5..\"}[5m])) / sum(rate(istio_requests_total[5m])) * 100"
          }
        ],
        "fieldConfig": {
          "defaults": {
            "thresholds": {
              "steps": [
                {"value": 0, "color": "green"},
                {"value": 1, "color": "yellow"},
                {"value": 5, "color": "red"}
              ]
            }
          }
        }
      },
      {
        "title": "P99 Latency",
        "type": "graph",
        "targets": [
          {
            "expr": "histogram_quantile(0.99, sum(rate(istio_request_duration_milliseconds_bucket{reporter=\"destination\"}[5m])) by (le, destination_service_name))",
            "legendFormat": "{{destination_service_name}}"
          }
        ]
      },
      {
        "title": "Service Topology",
        "type": "nodeGraph",
        "targets": [
          {
            "expr": "sum(rate(istio_requests_total{reporter=\"destination\"}[5m])) by (source_workload, destination_service_name)"
          }
        ]
      }
    ]
  }
}

模板 6：Kiali 服务网格可视化

# Kiali installation
apiVersion: kiali.io/v1alpha1
kind: Kiali
metadata:
  name: kiali
  namespace: istio-system
spec:
  auth:
    strategy: anonymous  # or openid, token
  deployment:
    accessible_namespaces:
      - "**"
  external_services:
    prometheus:
      url: http://prometheus.istio-system:9090
    tracing:
      url: http://jaeger-query.istio-system:16686
    grafana:
      url: http://grafana.istio-system:3000

模板 7：OpenTelemetry 集成

通过 OTLP 将数据发送到启用了 Jaeger OTLP 的接收器，并公开其 4317 Service 端口。标准发行版已移除原生 Jaeger Collector 导出器；请参阅 OpenTelemetry 迁移。下方的明文 insecure: true 示例仅适用于隔离网络；对于不可信链路，请使用经过身份验证的 TLS。请单独配置所引用的 Istio otel 扩展提供程序。

# OpenTelemetry Collector for mesh
apiVersion: v1
kind: ConfigMap
metadata:
  name: otel-collector-config
data:
  config.yaml: |
    receivers:
      otlp:
        protocols:
          grpc:
            endpoint: 0.0.0.0:4317
          http:
            endpoint: 0.0.0.0:4318
      zipkin:
        endpoint: 0.0.0.0:9411

    processors:
      batch:
        timeout: 10s

    exporters:
      otlp/jaeger:
        endpoint: jaeger-collector:4317
        tls:
          insecure: true
      prometheus:
        endpoint: 0.0.0.0:8889

    service:
      pipelines:
        traces:
          receivers: [otlp, zipkin]
          processors: [batch]
          exporters: [otlp/jaeger]
        metrics:
          receivers: [otlp]
          processors: [batch]
          exporters: [prometheus]
---
# Istio Telemetry v2 with OTel
apiVersion: telemetry.istio.io/v1alpha1
kind: Telemetry
metadata:
  name: mesh-default
  namespace: istio-system
spec:
  tracing:
    - providers:
        - name: otel
      randomSamplingPercentage: 10

告警规则

apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
  name: mesh-alerts
  namespace: istio-system
spec:
  groups:
    - name: mesh.rules
      rules:
        - alert: HighErrorRate
          expr: |
            sum(rate(istio_requests_total{response_code=~"5.."}[5m])) by (destination_service_name)
            / sum(rate(istio_requests_total[5m])) by (destination_service_name) > 0.05
          for: 5m
          labels:
            severity: critical
          annotations:
            summary: "High error rate for {{ $labels.destination_service_name }}"

        - alert: HighLatency
          expr: |
            histogram_quantile(0.99, sum(rate(istio_request_duration_milliseconds_bucket[5m]))
            by (le, destination_service_name)) > 1000
          for: 5m
          labels:
            severity: warning
          annotations:
            summary: "High P99 latency for {{ $labels.destination_service_name }}"

        - alert: MeshCertExpiring
          expr: |
            (certmanager_certificate_expiration_timestamp_seconds - time()) / 86400 < 7
          labels:
            severity: warning
          annotations:
            summary: "Mesh certificate expiring in less than 7 days"

最佳实践

应做事项

• 合理采样——开发环境中采样 100%，生产环境中采样 1-10%
• 使用追踪上下文——一致地传播请求头
• 设置告警——针对黄金信号设置告警
• 关联指标／追踪——使用 exemplars
• 合理保留数据——设置热／冷存储层级

不应做事项

• 不要过度采样——存储成本会不断累积
• 不要忽视基数——限制标签值
• 不要跳过仪表盘——可视化依赖关系
• 不要忽略成本——监控可观测性成本

资源

• Istio 可观测性
• Linkerd 可观测性
• OpenTelemetry
• Kiali

限制

• 仅当任务明确符合上述范围时，才使用此技能。
• 不要将输出视为环境特定验证、测试或专家审查的替代品。
• 如果缺少必要输入、权限、安全边界或成功标准，请暂停并请求澄清。