Kamixitong/scripts/setup_monitoring.py

#!/usr/bin/env python3
"""
监控告警配置脚本
自动安装和配置 Prometheus + Grafana 监控栈
"""
import os
import sys
import subprocess
import json
import time
from pathlib import Path


class MonitoringSetup:
    """监控配置类"""

    def __init__(self):
        self.monitoring_dir = Path('monitoring')
        self.docker_compose_file = self.monitoring_dir / 'docker-compose.yml'

    def check_docker(self):
        """检查 Docker 和 Docker Compose"""
        print("🔍 检查 Docker 环境...")

        try:
            result = subprocess.run(['docker', '--version'], capture_output=True, text=True)
            print(f"✅ Docker: {result.stdout.strip()}")
        except FileNotFoundError:
            print("❌ Docker 未安装")
            print("\n请安装 Docker: https://docs.docker.com/get-docker/")
            return False

        try:
            result = subprocess.run(['docker-compose', '--version'], capture_output=True, text=True)
            print(f"✅ Docker Compose: {result.stdout.strip()}")
        except FileNotFoundError:
            try:
                result = subprocess.run(['docker', 'compose', 'version'], capture_output=True, text=True)
                print(f"✅ Docker Compose: {result.stdout.strip()}")
            except FileNotFoundError:
                print("❌ Docker Compose 未安装")
                print("\n请安装 Docker Compose")
                return False

        return True

    def create_docker_compose(self):
        """创建 Docker Compose 配置"""
        print("\n📝 创建 Docker Compose 配置...")

        self.monitoring_dir.mkdir(exist_ok=True)

        compose_content = """version: '3.8'

services:
  # Prometheus 监控
  prometheus:
    image: prom/prometheus:latest
    container_name: kamaxitong-prometheus
    restart: unless-stopped
    ports:
      - "9090:9090"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
      - ./alert_rules.yml:/etc/prometheus/alert_rules.yml
      - prometheus_data:/prometheus
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
      - '--web.console.libraries=/etc/prometheus/console_libraries'
      - '--web.console.templates=/etc/prometheus/consoles'
      - '--storage.tsdb.retention.time=200h'
      - '--web.enable-lifecycle'
    networks:
      - monitoring

  # Grafana 仪表板
  grafana:
    image: grafana/grafana:latest
    container_name: kamaxitong-grafana
    restart: unless-stopped
    ports:
      - "3000:3000"
    volumes:
      - grafana_data:/var/lib/grafana
      - ./grafana/provisioning:/etc/grafana/provisioning
    environment:
      - GF_SECURITY_ADMIN_USER=admin
      - GF_SECURITY_ADMIN_PASSWORD=admin123
      - GF_USERS_ALLOW_SIGN_UP=false
    networks:
      - monitoring

  # Node Exporter 系统监控
  node-exporter:
    image: prom/node-exporter:latest
    container_name: kamaxitong-node-exporter
    restart: unless-stopped
    ports:
      - "9100:9100"
    volumes:
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /:/rootfs:ro
    command:
      - '--path.procfs=/host/proc'
      - '--path.rootfs=/rootfs'
      - '--path.sysfs=/host/sys'
      - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
    networks:
      - monitoring

  # AlertManager 告警管理
  alertmanager:
    image: prom/alertmanager:latest
    container_name: kamaxitong-alertmanager
    restart: unless-stopped
    ports:
      - "9093:9093"
    volumes:
      - ./alertmanager.yml:/etc/alertmanager/alertmanager.yml
      - alertmanager_data:/alertmanager
    command:
      - '--config.file=/etc/alertmanager/alertmanager.yml'
      - '--storage.path=/alertmanager'
    networks:
      - monitoring

  # Redis Exporter
  redis-exporter:
    image: oliver006/redis_exporter:latest
    container_name: kamaxitong-redis-exporter
    restart: unless-stopped
    ports:
      - "9121:9121"
    environment:
      - REDIS_ADDR=redis://redis:6379
    networks:
      - monitoring
    depends_on:
      - redis

  # Redis 数据库
  redis:
    image: redis:alpine
    container_name: kamaxitong-redis
    restart: unless-stopped
    ports:
      - "6379:6379"
    volumes:
      - redis_data:/data
    networks:
      - monitoring

volumes:
  prometheus_data:
  grafana_data:
  alertmanager_data:
  redis_data:

networks:
  monitoring:
    driver: bridge
"""

        with open(self.docker_compose_file, 'w') as f:
            f.write(compose_content)

        print(f"✅ Docker Compose 配置已创建: {self.docker_compose_file}")

    def create_alertmanager_config(self):
        """创建 AlertManager 配置"""
        print("\n📝 创建 AlertManager 配置...")

        alertmanager_config = self.monitoring_dir / 'alertmanager.yml'

        config_content = """global:
  smtp_smarthost: 'localhost:587'
  smtp_from: 'alerts@yourcompany.com'
  smtp_auth_username: 'alerts@yourcompany.com'
  smtp_auth_password: 'your-password'

route:
  group_by: ['alertname']
  group_wait: 10s
  group_interval: 10s
  repeat_interval: 1h
  receiver: 'web.hook'
  routes:
    - match:
        severity: critical
      receiver: 'critical-alerts'
    - match:
        severity: warning
      receiver: 'warning-alerts'

receivers:
  - name: 'web.hook'
    webhook_configs:
      - url: 'http://localhost:5001/alert'
        send_resolved: true

  - name: 'critical-alerts'
    email_configs:
      - to: 'admin@yourcompany.com'
        subject: '【严重告警】KaMiXiTong 系统告警'
        body: |
          {{ range .Alerts }}
          告警: {{ .Annotations.summary }}
          描述: {{ .Annotations.description }}
          时间: {{ .StartsAt }}
          级别: {{ .Labels.severity }}
          {{ end }}

  - name: 'warning-alerts'
    email_configs:
      - to: 'admin@yourcompany.com'
        subject: '【警告】KaMiXiTong 系统告警'
        body: |
          {{ range .Alerts }}
          告警: {{ .Annotations.summary }}
          描述: {{ .Annotations.description }}
          时间: {{ .StartsAt }}
          级别: {{ .Labels.severity }}
          {{ end }}

inhibit_rules:
  - source_match:
      severity: 'critical'
    target_match:
      severity: 'warning'
    equal: ['alertname', 'dev', 'instance']
"""

        with open(alertmanager_config, 'w') as f:
            f.write(config_content)

        print(f"✅ AlertManager 配置已创建: {alertmanager_config}")

    def create_grafana_provisioning(self):
        """创建 Grafana 配置"""
        print("\n📝 创建 Grafana 配置...")

        provisioning_dir = self.monitoring_dir / 'grafana' / 'provisioning'
        provisioning_dir.mkdir(parents=True, exist_ok=True)

        # 数据源配置
        datasource_config = provisioning_dir / 'datasources.yml'
        datasource_content = """apiVersion: 1

datasources:
  - name: Prometheus
    type: prometheus
    access: proxy
    url: http://prometheus:9090
    isDefault: true
"""

        with open(datasource_config, 'w') as f:
            f.write(datasource_content)

        # 仪表板配置
        dashboard_config = provisioning_dir / 'dashboards.yml'
        dashboard_content = """apiVersion: 1

providers:
  - name: 'KaMiXiTong'
    orgId: 1
    folder: 'KaMiXiTong'
    type: file
    disableDeletion: false
    editable: true
    options:
      path: /etc/grafana/provisioning/dashboards
"""

        with open(dashboard_config, 'w') as f:
            f.write(dashboard_content)

        print("✅ Grafana 配置已创建")

    def copy_monitoring_files(self):
        """复制监控配置文件"""
        print("\n📁 复制监控配置文件...")

        # 复制 Prometheus 配置
        prometheus_src = Path('monitoring/prometheus.yml')
        prometheus_dst = self.monitoring_dir / 'prometheus.yml'
        if prometheus_src.exists():
            prometheus_dst.write_text(prometheus_src.read_text())

        # 复制告警规则
        alert_rules_src = Path('monitoring/alert_rules.yml')
        alert_rules_dst = self.monitoring_dir / 'alert_rules.yml'
        if alert_rules_src.exists():
            alert_rules_dst.write_text(alert_rules_src.read_text())

        # 复制 Grafana 仪表板
        grafana_src = Path('monitoring/grafana_dashboard.json')
        grafana_dashboard_dir = self.monitoring_dir / 'grafana' / 'provisioning' / 'dashboards'
        grafana_dashboard_dir.mkdir(parents=True, exist_ok=True)
        grafana_dst = grafana_dashboard_dir / 'kamaxitong_dashboard.json'
        if grafana_src.exists():
            grafana_dst.write_text(grafana_src.read_text())

        print("✅ 配置文件已复制")

    def start_monitoring(self):
        """启动监控服务"""
        print("\n🚀 启动监控服务...")

        os.chdir(self.monitoring_dir)

        # 启动服务
        print("启动 Docker Compose...")
        result = subprocess.run(['docker-compose', 'up', '-d'], capture_output=True, text=True)

        if result.returncode != 0:
            print(f"❌ 启动失败: {result.stderr}")
            return False

        print("✅ 监控服务已启动")
        return True

    def show_access_info(self):
        """显示访问信息"""
        print("\n" + "=" * 60)
        print("✅ 监控服务配置完成!")
        print("=" * 60)
        print("\n📊 访问地址:")
        print("  Grafana 仪表板: http://localhost:3000")
        print("    用户名: admin")
        print("    密码: admin123")
        print("\n  Prometheus: http://localhost:9090")
        print("  AlertManager: http://localhost:9093")
        print("  Node Exporter: http://localhost:9100")
        print("\n📋 常用命令:")
        print("  查看服务状态: docker-compose ps")
        print("  查看日志: docker-compose logs -f")
        print("  停止服务: docker-compose down")
        print("  重启服务: docker-compose restart")
        print("\n⚠️  注意:")
        print("  - 首次启动 Grafana 需要导入仪表板")
        print("  - 定期备份 Grafana 数据")
        print("  - 配置邮件告警需要修改 alertmanager.yml")

    def run(self):
        """运行配置流程"""
        print("=" * 60)
        print("🔧 KaMiXiTong 监控告警配置工具")
        print("=" * 60)

        # 检查 Docker
        if not self.check_docker():
            sys.exit(1)

        # 创建配置
        self.create_docker_compose()
        self.create_alertmanager_config()
        self.create_grafana_provisioning()
        self.copy_monitoring_files()

        # 询问是否启动
        print("\n是否启动监控服务? (y/N)")
        if input().lower() == 'y':
            if self.start_monitoring():
                time.sleep(5)  # 等待服务启动
                self.show_access_info()
            else:
                print("❌ 启动失败，请检查错误信息")
                sys.exit(1)
        else:
            print("\n📋 手动启动命令:")
            print(f"  cd {self.monitoring_dir}")
            print("  docker-compose up -d")


def main():
    """主函数"""
    setup = MonitoringSetup()
    setup.run()


if __name__ == '__main__':
    main()