water-management-system/monitoring/alert_rules.yml

groups:
  - name: water-management-system
    rules:
      # 应用服务可用性告警
      - alert: ApplicationDown
        expr: up == 0
        for: 1m
        labels:
          severity: critical
        annotations:
          summary: "服务 {{ $labels.instance }} 已停止响应"
          description: "服务 {{ $labels.instance }} 在过去1分钟内没有响应"

      # 高CPU使用率告警
      - alert: HighCPUUsage
        expr: 100 - (avg by(instance) (rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100) > 80
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "高CPU使用率: {{ $labels.instance }}"
          description: "CPU使用率超过80%，持续5分钟"

      # 高内存使用率告警
      - alert: HighMemoryUsage
        expr: (1 - (node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes)) * 100 > 85
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "高内存使用率: {{ $labels.instance }}"
          description: "内存使用率超过85%，持续5分钟"

      # 数据库连接数告警
      - alert: HighDatabaseConnections
        expr: pg_stat_database_numbackends > 80
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "数据库连接数过高: {{ $labels.datname }}"
          description: "当前数据库连接数超过80个"

      # Redis内存使用率告警
      - alert: RedisMemoryUsage
        expr: redis_memory_used_bytes / redis_memory_max_bytes * 100 > 80
        for: 5m
        labels:
          severity: warning
        annotations:
          summary: "Redis内存使用率过高: {{ $labels.instance }}"
          description: "Redis内存使用率超过80%"