base/docker/prometheus/alerts.yml

groups:
  # ── 系统级告警 ──
  - name: system
    rules:
      - alert: HMSHighMemoryUsage
        expr: process_resident_memory_bytes > 800000000
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "HMS 内存使用超过 800MB"
          description: "当前值: {{ $value | humanize }}B"

      - alert: HMSHighMemoryCritical
        expr: process_resident_memory_bytes > 1000000000
        for: 2m
        labels:
          severity: critical
        annotations:
          summary: "HMS 内存使用超过 1GB（危险）"
          description: "当前值: {{ $value | humanize }}B"

      - alert: HMSHighCPU
        expr: rate(process_cpu_seconds_total[5m]) > 0.8
        for: 10m
        labels:
          severity: warning
        annotations:
          summary: "HMS CPU 使用率超过 80%"

  # ── 应用级告警 ──
  - name: application
    rules:
      - alert: HMSHighErrorRate
        expr: rate(http_requests_total{status=~"5.."}[5m]) / rate(http_requests_total[5m]) > 0.05
        for: 5m
        labels:
          severity: critical
        annotations:
          summary: "API 5xx 错误率超过 5%"
          description: "当前错误率: {{ $value | humanizePercentage }}"

      - alert: HMSSlowResponses
        expr: histogram_quantile(0.95, rate(http_request_duration_seconds_bucket[5m])) > 2
        for: 10m
        labels:
          severity: warning
        annotations:
          summary: "95% 请求响应时间超过 2 秒"

      - alert: HMSInstanceDown
        expr: up{job="hms"} == 0
        for: 2m
        labels:
          severity: critical
        annotations:
          summary: "HMS 服务不可达"

  # ── 数据库告警 ──
  - name: database
    rules:
      - alert: HMSPostgresConnectionsHigh
        expr: pg_stat_activity_count > 80
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "PostgreSQL 活跃连接数超过 80"

      - alert: HMSPostgresReplicationLag
        expr: pg_replication_lag > 30
        for: 5m
        labels:
          severity: critical
        annotations:
          summary: "PostgreSQL 复制延迟超过 30 秒"

      - alert: HMSBackupMissing
        expr: time() - hms_last_backup_timestamp > 86400 * 2
        for: 1h
        labels:
          severity: critical
        annotations:
          summary: "数据库备份超过 48 小时未执行"

  # ── Redis 告警 ──
  - name: redis
    rules:
      - alert: HMSRedisMemoryHigh
        expr: redis_memory_used_bytes / redis_memory_max_bytes > 0.9
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "Redis 内存使用超过 90%"

      - alert: HMSRedisDown
        expr: redis_up == 0
        for: 2m
        labels:
          severity: critical
        annotations:
          summary: "Redis 服务不可达"