1
# Helm values.yaml
2
higress-core:
3
  gateway:
4
    metrics:
5
      enabled: true
6
      port: 15020          # Metrics 端口
7
      path: /stats/prometheus
8

9
# 或手动配置
10
apiVersion: v1
11
kind: ConfigMap
12
metadata:
13
  name: higress-config
14
  namespace: higress-system
15
data:
16
  higress: |
17
    o11y:
18
      metrics:
19
        enabled: true
20
        port: 15020

1
# QPS（每秒请求数）
2
rate(envoy_http_downstream_rq_total{pod=~"higress-gateway.*"}[1m])
3

4
# P99 延迟
5
histogram_quantile(0.99,
6
  rate(envoy_http_downstream_rq_time_bucket{pod=~"higress-gateway.*"}[5m])
7
)
8

9
# 错误率
10
sum(rate(envoy_http_downstream_rq_5xx[5m]))
11
/
12
sum(rate(envoy_http_downstream_rq_total[5m]))
13

14
# 上游健康度
15
envoy_cluster_upstream_cx_active{cluster_name=~".*"}

1
# Prometheus AlertManager 规则
2
groups:
3
  - name: higress-alerts
4
    rules:
5
      # 高错误率告警
6
      - alert: HigressHighErrorRate
7
        expr: |
8
          sum(rate(envoy_http_downstream_rq_5xx[5m]))
9
          /
10
          sum(rate(envoy_http_downstream_rq_total[5m])) > 0.05
11
        for: 2m
12
        labels:
13
          severity: critical
14
        annotations:
15
          summary: "Higress 错误率超过 5%"
16

17
      # 高延迟告警
18
      - alert: HigressHighLatency
19
        expr: |
20
          histogram_quantile(0.99,
21
            rate(envoy_http_downstream_rq_time_bucket[5m])
22
          ) > 1000
23
        for: 5m
24
        labels:
25
          severity: warning
26
        annotations:
27
          summary: "Higress P99 延迟超过 1s"
28

29
      # 上游服务不可用
30
      - alert: HigressUpstreamDown
31
        expr: envoy_cluster_upstream_cx_active == 0
32
        for: 1m
33
        labels:
34
          severity: critical

1
apiVersion: v1
2
kind: ConfigMap
3
metadata:
4
  name: higress-config
5
  namespace: higress-system
6
data:
7
  higress: |
8
    o11y:
9
      tracing:
10
        enabled: true
11
        # Jaeger 配置
12
        provider: jaeger
13
        jaeger:
14
          collector: jaeger-collector.jaeger.svc.cluster.local:9411
15
          # 采样率（0.0-1.0）
16
          sampling:
17
            type: probabilistic
18
            value: 0.1        # 10% 采样

1
data:
2
  higress: |
3
    o11y:
4
      tracing:
5
        enabled: true
6
        provider: skywalking
7
        skywalking:
8
          service: oap.skywalking.svc.cluster.local:11800
9
          sampling:
10
            type: rate
11
            value: 100        # 每秒 100 个请求

1
# Higress 自动传播 Trace Header
2
# 支持的 Header：
3
# - X-Request-Id（Envoy 原生）
4
# - X-B3-TraceId / X-B3-SpanId（Zipkin）
5
# - traceparent / tracestate（W3C Trace Context）
6
# - sw8（SkyWalking）
7

8
# 上游服务需要：
9
# 1. 读取这些 Header
10
# 2. 创建子 Span
11
# 3. 继续传播到下游

1
apiVersion: v1
2
kind: ConfigMap
3
metadata:
4
  name: higress-config
5
  namespace: higress-system
6
data:
7
  higress: |
8
    accessLog:
9
      enabled: true
10
      # JSON 格式（推荐，便于 ELK 解析）
11
      format: json
12
      # 自定义字段
13
      template: |
14
        {
15
          "timestamp": "%START_TIME%",
16
          "method": "%REQ(:METHOD)%",
17
          "path": "%REQ(X-ENVOY-ORIGINAL-PATH?:PATH)%",
18
          "protocol": "%PROTOCOL%",
19
          "response_code": "%RESPONSE_CODE%",
20
          "response_flags": "%RESPONSE_FLAGS%",
21
          "bytes_received": "%BYTES_RECEIVED%",
22
          "bytes_sent": "%BYTES_SENT%",
23
          "duration": "%DURATION%",
24
          "upstream_service_time": "%RESP(X-ENVOY-UPSTREAM-SERVICE-TIME)%",
25
          "x_forwarded_for": "%REQ(X-FORWARDED-FOR)%",
26
          "user_agent": "%REQ(USER-AGENT)%",
27
          "request_id": "%REQ(X-REQUEST-ID)%",
28
          "authority": "%REQ(:AUTHORITY)%",
29
          "upstream_host": "%UPSTREAM_HOST%",
30
          "upstream_cluster": "%UPSTREAM_CLUSTER%",
31
          "route_name": "%ROUTE_NAME%"
32
        }

1
Higress Gateway
2
   ├── Metrics → Prometheus → Grafana（监控面板）
3
   ├── Tracing → Jaeger/SkyWalking（链路追踪）
4
   └── Logging → Fluentd → Elasticsearch → Kibana（日志分析）

1
higress-core:
2
  gateway:
3
    metrics:
4
      enabled: true
5
      port: 15020
6

7
    accessLog:
8
      enabled: true
9
      format: json
10

11
global:
12
  o11y:
13
    enabled: true
14
    grafana:
15
      enabled: true
16
      # 自动导入 Higress Dashboard
17
      dashboards:
18
        higress: true
19
    prometheus:
20
      enabled: true
21
      # 自动配置 scrape
22
      scrapeHigress: true
23
    tracing:
24
      enabled: true
25
      provider: jaeger
26
      jaeger:
27
        collector: jaeger-collector.jaeger.svc.cluster.local:9411
28
        sampling:
29
          type: probabilistic
30
          value: 0.1

1
# 1. 降低采样率（只记录错误和慢请求）
2
accessLog:
3
  enabled: true
4
  # 只记录 5xx 和 >1s 的请求
5
  filter:
6
    statusCode: ">=500"
7
    duration: ">=1000"
8

9
# 2. 精简字段
10
template: |
11
  {
12
    "timestamp": "%START_TIME%",
13
    "method": "%REQ(:METHOD)%",
14
    "path": "%REQ(X-ENVOY-ORIGINAL-PATH?:PATH)%",
15
    "response_code": "%RESPONSE_CODE%",
16
    "duration": "%DURATION%"
17
  }
18

19
# 3. 日志轮转
20
# 配置 logrotate 或 Fluentd buffer