API 网关监控三件套保姆级教程：Prometheus + Loki + Grafana 从零搭建，30 分钟搞定可观测性！
8/30/2026

文章链接：https://blog.csdn.net/qq_43201350/article/details/163477763

“网关上线第二天，凌晨三点被报警电话吵醒——上游服务挂了，网关返回 502，但没人知道是因为什么。”

这是我三年前的经历。从那以后我学会了一件事： 网关可观测性不是选配，是刚需。 今天这篇就带你从零搭建一套生产级监控体系，30 分钟内搞定。

一、可观测性三要素

在网关场景下，我们关心的是：

维度核心问题对应工具
指标（Metrics）QPS 多高？延迟多大？错误率多少？Prometheus
日志（Logs）哪个请求报错了？具体原因？Loki / ELK
追踪（Traces）一次请求穿过网关 → 后端 → DB 的完整链路Jaeger / SkyWalking
可视化把所有数据变成图表和大盘Grafana

本文聚焦最核心的三件套： Prometheus（采集） + Loki（日志） + Grafana（展示） 。

为什么选这套？因为它们是 CNCF 毕业项目，社区最活跃，Docker 一键部署。

二、整体架构

┌────────────────────────────────────────────────────────┐
│                     Grafana (端口 3000)                  │
│   ┌──────────┐  ┌──────────┐  ┌──────────────────┐    │
│   │ 网关大盘  │  │ 日志面板  │  │  告警规则         │    │
│   └──────────┘  └──────────┘  └──────────────────┘    │
└──────────┬──────────────────┬──────────────────────────┘
           │ 查指标            │ 查日志
           ▼                   ▼
┌──────────────────┐  ┌──────────────────┐
│   Prometheus     │  │      Loki        │
│   (端口 9090)    │  │   (端口 3100)    │
│                  │  │                  │
│  ┌────────────┐  │  │  ┌────────────┐  │
│  │ 指标存储    │  │  │  │ 日志存储    │  │
│  │ (TSDB)     │  │  │  │ (Index+Chunk)│ │
│  └────────────┘  │  │  └────────────┘  │
└────────▲─────────┘  └────────▲─────────┘
         │ 抓取 /metrics    │ 推送日志
         │                   │
┌────────┴───────────────────┴─────────────────────────┐
│                   API 网关 (端口 9080)                  │
│  ┌─────────────────────────────────────────────────┐  │
│  │  prometheus 插件 (暴露 /apisix/prometheus/metrics) │  │
│  │  http-logger 插件 (推送日志到 Loki)               │  │
│  └─────────────────────────────────────────────────┘  │
└───────────────────────────────────────────────────────┘

三、环境搭建

3.1 docker-compose.yml

在之前的基础上加入监控组件：

version: '3.8'
 
services:
  # =========== 配置中心 ===========
  etcd:
    image: bitnami/etcd:3.5
    environment:
      - ALLOW_NONE_AUTHENTICATION=yes
      - ETCD_ADVERTISE_CLIENT_URLS=http://0.0.0.0:2379
    ports:
      - "2379:2379"
    networks:
      - demo-net
 
  # =========== API 网关 ===========
  gateway:
    image: apache/apisix:3.10.0-centos
    depends_on:
      - etcd
    volumes:
      - ./config/gateway-config.yaml:/usr/local/apisix/conf/config.yaml
    ports:
      - "9080:9080"
      - "9180:9180"
    environment:
      - APISIX_STAND_ALONE=false
    networks:
      - demo-net
 
  # =========== 示例后端服务 ===========
  backend-v1:
    image: ealen/echo-server:latest
    environment:
      - PORT=8080
    networks:
      - demo-net
 
  backend-v2:
    image: ealen/echo-server:latest
    environment:
      - PORT=8080
    networks:
      - demo-net
 
  # =========== Prometheus ===========
  prometheus:
    image: prom/prometheus:v2.52.0
    volumes:
      - ./prometheus/prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus_data:/prometheus
    ports:
      - "9090:9090"
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
      - '--storage.tsdb.retention.time=15d'
      - '--web.enable-lifecycle'
    networks:
      - demo-net
 
  # =========== Loki ===========
  loki:
    image: grafana/loki:2.9.0
    ports:
      - "3100:3100"
    command: -config.file=/etc/loki/local-config.yaml
    volumes:
      - loki_data:/loki
    networks:
      - demo-net
 
  # =========== Grafana ===========
  grafana:
    image: grafana/grafana:10.4.0
    ports:
      - "3000:3000"
    environment:
      - GF_SECURITY_ADMIN_USER=demo_admin
      - GF_SECURITY_ADMIN_PASSWORD=demo_admin_2026
      - GF_AUTH_ANONYMOUS_ENABLED=false
      - GF_INSTALL_PLUGINS=grafana-piechart-panel
    volumes:
      - grafana_data:/var/lib/grafana
      - ./grafana/provisioning:/etc/grafana/provisioning
    networks:
      - demo-net
    depends_on:
      - prometheus
      - loki
 
  # =========== 压测工具 ===========
  loadgen:
    image: alpine/curl:latest
    entrypoint: ["/bin/sh", "-c", "while true; do sleep 3600; done"]
    networks:
      - demo-net
 
networks:
  demo-net:
    driver: bridge
 
volumes:
  prometheus_data:
  loki_data:
  grafana_data:

3.2Prometheus配置

# prometheus/prometheus.yml
global:
  scrape_interval: 15s
  evaluation_interval: 15s
  external_labels:
    cluster: 'demo-gateway-cluster'
    env: 'demo'
 
scrape_configs:
  # === 抓取网关指标 ===
  - job_name: 'api-gateway'
    metrics_path: '/apisix/prometheus/metrics'
    static_configs:
      - targets: ['gateway:9091']
        labels:
          service: 'api-gateway'
          instance: 'gateway-demo-01'
 
  # === Prometheus 自监控 ===
  - job_name: 'prometheus'
    static_configs:
      - targets: ['localhost:9090']

3.3Grafana数据源预配置

# grafana/provisioning/datasources/datasources.yaml
apiVersion: 1
 
datasources:
  - name: Prometheus
    type: prometheus
    access: proxy
    url: http://prometheus:9090
    isDefault: true
    editable: true
 
  - name: Loki
    type: loki
    access: proxy
    url: http://loki:3100
    editable: true
# grafana/provisioning/dashboards/dashboards.yaml
apiVersion: 1
 
providers:
  - name: 'Gateway Dashboards'
    orgId: 1
    folder: 'API Gateway'
    type: file
    disableDeletion: false
    updateIntervalSeconds: 30
    allowUiUpdates: true
    options:
      path: /etc/grafana/provisioning/dashboards

四、启用网关指标暴露

4.1 配置 prometheus 插件

# 全局启用 prometheus 插件
curl -s http://127.0.0.1:9180/admin-api/global_rules -X PUT -d '
{
  "plugins": {
    "prometheus": {
      "prefer_name": true
    }
  }
}'

配置后网关会在 http://gateway:9091/apisix/prometheus/metrics 暴露指标。

4.2 核心指标一览

# 拉取指标看看有哪些
curl -s http://127.0.0.1:9091/apisix/prometheus/metrics | grep -E "^apisix_" | head -30

关键指标分类：

# ─── HTTP 请求量 ───
apisix_http_status{code="200"}     # 按状态码统计
apisix_http_requests_total         # 请求总量
 
# ─── 延迟分布 ───
apisix_http_latency_bucket         # 延迟直方图
apisix_http_latency_sum            # 总延迟
 
# ─── 带宽 ───
apisix_bandwidth_total             # 总带宽
apisix_bandwidth{type="ingress"}   # 入向
apisix_bandwidth{type="egress"}    # 出向
 
# ─── 连接数 ───
apisix_nginx_http_current_connections  # 当前连接数
 
# ─── etcd 健康 ───
apisix_etcd_modify_indexes         # etcd 索引版本
apisix_etcd_reachable              # etcd 可达性 (1=正常)

4.3 模拟流量——让图表有数据

# 先在网关配一条测试路由
curl -s http://127.0.0.1:9180/admin-api/upstreams/demo-echo-upstream -X PUT -d '
{
  "type": "roundrobin",
  "nodes": {
    "backend-v1:8080": 1,
    "backend-v2:8080": 1
  }
}'
 
curl -s http://127.0.0.1:9180/admin-api/routes/demo-echo-route -X PUT -d '
{
  "uri": "/demo/echo/*",
  "upstream_id": "demo-echo-upstream",
  "plugins": {
    "prometheus": {"prefer_name": false}
  }
}'
# 模拟持续流量——在 docker 容器中跑
docker exec -d $(docker ps -qf "name=loadgen") sh -c '
while true; do
  # 正常请求 - 80%
  for i in $(seq 1 8); do
    curl -s -o /dev/null -w "%{http_code}\n" \
      http://gateway:9080/demo/echo/hello &
  done
 
  # 慢请求 - 10%
  curl -s -o /dev/null -w "%{http_code}\n" \
    http://gateway:9080/demo/echo/slow &
 
  # 错误请求 - 5%
  curl -s -o /dev/null -w "%{http_code}\n" \
    http://gateway:9080/demo/echo/not-found &
 
  # 超时请求 - 5%
  curl -s -o /dev/null --max-time 1 \
    http://gateway:9080/demo/echo/timeout &
 
  sleep 2
done
'

五、搭建 Grafana 大盘

5.1 第一个面板：QPS & 错误率

登录 Grafana（ http://localhost:3000 ，账号 demo_admin ），新建 Dashboard。

面板 1：每秒请求量（QPS）

类型: Graph (Time Series)
PromQL: sum(rate(apisix_http_requests_total[1m])) by (route)
Legend: {{route}}

面板 2：HTTP 状态码分布

类型: Pie Chart
PromQL: sum(rate(apisix_http_status[1m])) by (code)
Legend: {{code}}

面板 3：错误率趋势

类型: Stat
PromQL: 
  sum(rate(apisix_http_status{code=~"5.."}[5m])) 
  / 
  sum(rate(apisix_http_status[5m])) * 100
Unit: percent (0-100)
Thresholds: 绿色 < 1%, 黄色 < 5%, 红色 >= 5%

面板 4：P99 延迟

类型: Graph
PromQL: histogram_quantile(0.99, sum(rate(apisix_http_latency_bucket[5m])) by (le, route))
Legend: P99 - {{route}}

5.2 第二个面板：上游健康度

# 上游请求成功率
sum(rate(apisix_http_status{code!~"5.."}[5m])) by (node)
/
sum(rate(apisix_http_status[5m])) by (node) * 100
 
# 上游平均延迟
sum(rate(apisix_http_latency_sum[5m])) by (node)
/
sum(rate(apisix_http_latency_count[5m])) by (node) * 1000

5.3 组合成完整大盘 JSON

{
  "dashboard": {
    "title": "API 网关监控大盘 (Demo)",
    "uid": "demo-gateway-dashboard",
    "tags": ["gateway", "api", "demo"],
    "timezone": "browser",
    "panels": [
      {
        "title": "QPS (每秒请求)",
        "type": "timeseries",
        "targets": [{
          "expr": "sum(rate(apisix_http_requests_total[1m])) by (route)",
          "legendFormat": "{{route}}"
        }],
        "gridPos": {"x": 0, "y": 0, "w": 12, "h": 8}
      },
      {
        "title": "状态码分布率",
        "type": "stat",
        "targets": [
          {"expr": "sum(rate(apisix_http_status{code=\"200\"}[1m]))", "legendFormat": "2xx"},
          {"expr": "sum(rate(apisix_http_status{code=~\"4..\"}[1m]))", "legendFormat": "4xx"},
          {"expr": "sum(rate(apisix_http_status{code=~\"5..\"}[1m]))", "legendFormat": "5xx"}
        ],
        "gridPos": {"x": 12, "y": 0, "w": 6, "h": 8},
        "fieldConfig": {
          "defaults": {
            "color": {"mode": "thresholds"},
            "thresholds": {
              "steps": [
                {"value": null, "color": "green"},
                {"value": 10, "color": "orange"},
                {"value": 50, "color": "red"}
              ]
            }
          }
        }
      },
      {
        "title": "P50 / P99 延迟 (ms)",
        "type": "timeseries",
        "targets": [
          {
            "expr": "histogram_quantile(0.50, sum(rate(apisix_http_latency_bucket[5m])) by (le)) * 1000",
            "legendFormat": "P50"
          },
          {
            "expr": "histogram_quantile(0.99, sum(rate(apisix_http_latency_bucket[5m])) by (le)) * 1000",
            "legendFormat": "P99"
          }
        ],
        "gridPos": {"x": 0, "y": 8, "w": 12, "h": 8}
      },
      {
        "title": "上游服务健康度",
        "type": "timeseries",
        "targets": [{
          "expr": "sum(rate(apisix_http_status{code!~\"5..\"}[5m])) by (node) / sum(rate(apisix_http_status[5m])) by (node) * 100",
          "legendFormat": "{{node}} 成功率"
        }],
        "gridPos": {"x": 12, "y": 8, "w": 6, "h": 8},
        "fieldConfig": {
          "defaults": {
            "unit": "percent",
            "min": 0,
            "max": 100
          }
        }
      },
      {
        "title": "当前活跃连接数",
        "type": "gauge",
        "targets": [{
          "expr": "apisix_nginx_http_current_connections",
          "legendFormat": "活跃连接"
        }],
        "gridPos": {"x": 0, "y": 16, "w": 6, "h": 8},
        "fieldConfig": {
          "defaults": {
            "thresholds": {
              "steps": [
                {"value": null, "color": "green"},
                {"value": 1000, "color": "orange"},
                {"value": 5000, "color": "red"}
              ]
            }
          }
        }
      },
      {
        "title": "etcd 健康检查",
        "type": "stat",
        "targets": [{
          "expr": "apisix_etcd_reachable",
          "legendFormat": "etcd 可达"
        }],
        "gridPos": {"x": 6, "y": 16, "w": 6, "h": 8},
        "fieldConfig": {
          "defaults": {
            "thresholds": {
              "steps": [
                {"value": null, "color": "red"},
                {"value": 1, "color": "green"}
              ]
            }
          }
        }
      },
      {
        "title": "带宽趋势 (入/出)",
        "type": "timeseries",
        "targets": [
          {
            "expr": "rate(apisix_bandwidth{type=\"ingress\"}[5m])",
            "legendFormat": "入向"
          },
          {
            "expr": "rate(apisix_bandwidth{type=\"egress\"}[5m])",
            "legendFormat": "出向"
          }
        ],
        "gridPos": {"x": 12, "y": 16, "w": 6, "h": 8}
      }
    ]
  }
}

5.4 导入大盘

将上面的 JSON 导入 Grafana：

1. 登录 Grafana → 左侧 "+" → Import
2. 粘贴 JSON → Load
3. 选择 Prometheus 数据源 → Import

效果概览：

区域面板说明
左上QPS 曲线实时请求速率
右上状态码计数2xx/4xx/5xx 分类
左中P50/P99 延迟性能瓶颈一目了然
右中上游健康度哪个节点出问题立刻发现
左下连接数仪表盘并发压力指示
中下etcd 健康配置中心是否在线
右下带宽趋势入/出流量监控

六、网关日志推送到 Loki

6.1 为什么要单独做日志

Prometheus 指标告诉你有多少错误，但 Loki 日志告诉你具体错在哪一行 。

6.2 配置日志推送

# 在全局规则中增加 http-logger
curl -s http://127.0.0.1:9180/admin-api/global_rules -X PATCH -d '
{
  "plugins": {
    "prometheus": {
      "prefer_name": true
    },
    "http-logger": {
      "uri": "http://loki:3100/loki/api/v1/push",
      "batch_max_size": 100,
      "max_retry_count": 3,
      "retry_delay": 1,
      "inactive_timeout": 5,
      "concat_method": "new_line",
      "include_req_body": false,
      "include_resp_body": false
    }
  }
}'

6.3 Loki 日志查询

在 Grafana 中切换到 Explore 页面，选择 Loki 数据源：

# 查询所有 5xx 错误的日志
{job="api-gateway"} | json | response_status >= 500
 
# 查询慢请求（延迟 > 1s）
{job="api-gateway"} | json | latency > 1
 
# 查询特定路由的错误
{job="api-gateway"} | json | route = "demo-echo-route" | response_status >= 500
 
# 查询近 5 分钟的 502 错误
{job="api-gateway"} 
| json 
| response_status = 502 
| line_format "时间: {{.timestamp}}, 上游: {{.upstream}}, 原因: {{.error_msg}}"

6.4 排障实战：模拟故障

# 步骤一：干掉一个上游节点
docker stop $(docker ps -qf "name=backend-v2")
 
# 步骤二：观察 Grafana
# → "上游服务健康度" 面板中 backend-v2:8080 成功率会从 ~100% 掉到 0%
# → "状态码分布" 面板中 502 计数会上升
# → Loki 中会看到具体的 502 日志
 
# 步骤三：在 Grafana → Explore → Loki 中查询
# {job="api-gateway"} | json | response_status = 502
# 会看到类似：
# upstream: "backend-v2:8080", error: "connection refused"

这就是 指标先报警、日志后定位 的经典排障流程。

七、告警规则配置

光看大盘不够，还要配告警——没人会 24 小时盯着屏幕。

7.1 关键告警规则

# prometheus/rules/gateway-alerts.yml
groups:
  - name: api_gateway_alerts
    interval: 30s
    rules:
      # ===== 可用性告警 =====
      - alert: HighErrorRate
        expr: |
          sum(rate(apisix_http_status{code=~"5.."}[5m])) 
          / 
          sum(rate(apisix_http_status[5m])) > 0.05
        for: 2m
        labels:
          severity: critical
        annotations:
          summary: "网关 5xx 错误率超过 5%"
          description: "当前值: {{ $value | humanizePercentage }}，路由: {{ $labels.route }}"
 
      - alert: UpstreamDown
        expr: |
          sum(rate(apisix_http_status{code=~"5.."}[5m])) by (node)
          / 
          sum(rate(apisix_http_status[5m])) by (node) > 0.5
        for: 1m
        labels:
          severity: critical
        annotations:
          summary: "上游节点 {{ $labels.node }} 不可用"
          description: "5xx 比例超过 50%，请检查上游服务"
 
      # ===== 性能告警 =====
      - alert: HighLatency
        expr: |
          histogram_quantile(0.99,
            sum(rate(apisix_http_latency_bucket[5m])) by (le, route)
          ) > 3
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "P99 延迟超过 3 秒"
          description: "路由 {{ $labels.route }}，当前值: {{ $value }}s"
 
      - alert: HighConnectionCount
        expr: apisix_nginx_http_current_connections > 5000
        for: 3m
        labels:
          severity: warning
        annotations:
          summary: "网关活跃连接数超过 5000"
          description: "当前连接数: {{ $value }}"
 
      # ===== 基础设施告警 =====
      - alert: EtcdUnreachable
        expr: apisix_etcd_reachable != 1
        for: 1m
        labels:
          severity: critical
        annotations:
          summary: "网关无法连接 etcd"
          description: "网关配置可能无法热更新，请立即检查"
 
      - alert: NoTraffic
        expr: sum(rate(apisix_http_requests_total[5m])) == 0
        for: 10m
        labels:
          severity: warning
        annotations:
          summary: "网关流量为零"
          description: "近 10 分钟无请求，可能配置异常或上游全部故障"

7.2 注册告警规则

# 在 prometheus.yml 中追加
rule_files:
  - '/etc/prometheus/rules/gateway-alerts.yml'
# 重新加载 Prometheus 配置
curl -X POST http://127.0.0.1:9090/-/reload

7.3 告警通知渠道

Prometheus 的Alertmanager负责把告警推送到即时通讯/邮件：

# alertmanager/config.yml
route:
  group_by: ['alertname', 'severity']
  group_wait: 10s
  group_interval: 10s
  repeat_interval: 1h
 
  # 严重告警走即时通讯
  routes:
    - match:
        severity: critical
      receiver: 'dingtalk-critical'
    - match:
        severity: warning
      receiver: 'dingtalk-warning'
 
receivers:
  - name: 'dingtalk-critical'
    webhook_configs:
      - url: 'https://oapi.dingtalk.com/robot/send?access_token=YOUR_TOKEN'
        send_resolved: true
        message: |
          ### 🔴 严重告警
          - 告警: {{ .CommonAnnotations.summary }}
          - 详情: {{ .CommonAnnotations.description }}
 
  - name: 'dingtalk-warning'
    webhook_configs:
      - url: 'https://oapi.dingtalk.com/robot/send?access_token=YOUR_TOKEN'
        send_resolved: true

八、一键部署脚本

把以上所有内容总结成一个脚本：

#!/bin/bash
# scripts/deploy-observability.sh
# 一键部署：网关 + Prometheus + Loki + Grafana
 
set -e
 
echo "========================================="
echo "  网关可观测性三件套 - 一键部署"
echo "========================================="
 
# 1. 创建目录结构
mkdir -p prometheus/rules grafana/provisioning/datasources grafana/provisioning/dashboards config
 
# 2. 检查 Docker
if ! command -v docker &> /dev/null; then
    echo "❌ 请先安装 Docker"
    exit 1
fi
 
# 3. 启动全部服务
docker compose up -d
 
# 4. 等待网关就绪
echo "⏳ 等待网关就绪..."
for i in $(seq 1 30); do
    if curl -s http://127.0.0.1:9180/admin-api/routes > /dev/null 2>&1; then
        echo "✅ 网关已就绪"
        break
    fi
    sleep 2
done
 
# 5. 配置网关路由
echo "📝 配置网关路由和监控插件..."
bash scripts/setup-gateway-routes.sh
 
# 6. 检查组件状态
echo ""
echo "========================================="
echo "  部署完成！访问地址："
echo "========================================="
echo "  Grafana:     http://localhost:3000"
echo "    账号: demo_admin"
echo "    密码: demo_admin_2026"
echo "  Prometheus:  http://localhost:9090"
echo "  网关 Admin:  http://localhost:9180"
echo "  网关 Proxy:  http://localhost:9080"
echo ""
echo "  下一步："
echo "  1. 浏览器打开 http://localhost:3000"
echo "  2. 左侧 → Dashboards → API 网关监控大盘"
echo "  3. 运行 scripts/simulate-traffic.sh 产生测试数据"
echo "========================================="

九、总结

9.1 我们搭建了什么

组件作用端口
Prometheus采集指标、存储时序数据9090
Loki采集日志、支持类似 LogQL 查询3100
Grafana统一可视化大盘 + 告警展示3000
网关 prometheus 插件暴露请求量/延迟/状态码/带宽自动
网关 http-logger 插件推送请求日志到 Loki自动

9.2 监控检查清单

上线前确认以下内容全部就绪：

  • Prometheus 能抓到网关指标（ up = 1）

  • Grafana 大盘 7 个面板都有数据

  • Loki 能查到网关日志

  • 告警规则已注册（在 Prometheus → Alerts 中可见）

  • 手动停一个上游节点，确认大盘能反映

  • 确认 Alertmanager 能发送测试告警

9.3 进阶方向

方向工具说明
链路追踪Jaeger / Zipkin看一次请求从网关到后端的完整路径
SLO 监控Pyrra / Sloth基于 SLI 指标，计算 burn rate
成本分析OpenCost网关带宽成本归属
异常检测基于 Z-score 的 PromQL自动发现异常流量模式

写在最后：监控不是终点，可观测性才是

恭喜你跟着这篇教程把 Prometheus + Loki + Grafana 跑起来了。

但说实话，"能装上"和"能对客户承诺 SLA"之间，差了十万八千里。

你可能很快会遇到这些问题：

  • 🔸 告警太多，每天收到 200 条，根本分不清哪个重要——告警风暴怎么治？

  • 🔸 阈值设多少合适？QPS 报警设 1000 还是 2000？拍脑袋设的阈值要么误报要么漏报

  • 🔸 凌晨出故障，谁响应？多久响应？怎么升级？——没有 On-Call SOP 就是裸奔

  • 🔸 Prometheus 自己挂了谁监控 Prometheus？——监控系统的监控盲区

  • 🔸 怎么证明你的系统达到了 99.9% 可用？——SLA 指标体系怎么设计

这些问题，我在付费专栏里用一整篇文章做了系统解答，包含：

  • SLA 指标体系设计（USE 方法论）

  • 告警分级标准（P0-P3）+ 阈值制定方法论

  • On-Call 排班 SOP 模板

  • 故障演练方案（定期注入故障，验证监控是否真的能发现问题）

  • 监控自身的监控方案

🎯 专栏《云原生API网关从入门到生产》完整包含：

  • ✅ 10 篇实战教程 + 8 项增值内容（代码仓库 / Checklist / 事故手册 / 面试题……）

  • ✅ Grafana Dashboard JSON（导入即用，含 SLA / 延迟 / 错误率 / 流量四块大盘）

  • ✅ 告警规则完整 YAML 合集

  • ✅ 付费读者优先答疑

👉 点击订阅《云原生 API 网关从入门到生产》19.9 元

这篇免费教程如果帮到了你，点赞收藏是对我最好的鼓励 🙏 我最好的鼓励 🙏

AI可观测性