Nginx + Docker + Kubernetes
在企业级 AI Agent 系统中的应用

# Nginx AI Agent 网关配置示例
worker_processes auto;
worker_rlimit_nofile 65535;

events {
    worker_connections 4096;
    use epoll;
    multi_accept on;
}

http {
    # 性能优化
    sendfile on;
    tcp_nopush on;
    tcp_nodelay on;
    keepalive_timeout 65;
    
    # AI 请求特殊配置
    client_max_body_size 100M;  # 支持大文件上传
    proxy_read_timeout 300s;     # AI 推理超时时间
    
    # 上游服务器组 - AI Agent 服务
    upstream ai_agent_backend {
        least_conn;  # 最少连接负载均衡
        server agent-service-1:8080 weight=3;
        server agent-service-2:8080 weight=3;
        server agent-service-3:8080 weight=2 backup;
        keepalive 32;
    }
    
    server {
        listen 443 ssl http2;
        server_name ai-agent.example.com;
        
        # SSL 配置
        ssl_certificate /etc/nginx/ssl/cert.pem;
        ssl_certificate_key /etc/nginx/ssl/key.pem;
        
        # AI Agent API 路由
        location /api/v1/agent/ {
            proxy_pass http://ai_agent_backend;
            proxy_http_version 1.1;
            proxy_set_header Host $host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            
            # 限流配置
            limit_req zone=ai_api burst=20 nodelay;
        }
    }
}

# 限流区域定义
limit_req_zone $binary_remote_addr zone=ai_api:10m rate=10r/s;
limit_req_zone $server_name zone=ai_global:10m rate=1000r/s;

# 熔断配置（使用 nginx-upstream-check-module）
upstream ai_agent_backend {
    server agent-1:8080;
    server agent-2:8080;
    check interval=3000 rise=2 fall=5 timeout=1000 type=http;
    check_http_send "HEAD /health HTTP/1.0\r\n\r\n";
    check_http_expect_alive http_2xx;
}

# 多阶段构建 - 减小镜像体积
FROM nvidia/cuda:12.1.0-cudnn8-runtime-ubuntu22.04 AS base

# 安装 Python 和依赖
RUN apt-get update && apt-get install -y \
    python3.10 \
    python3-pip \
    && rm -rf /var/lib/apt/lists/*

WORKDIR /app

# 复制依赖文件
COPY requirements.txt .

# 安装 Python 依赖（使用国内镜像加速）
RUN pip3 install --no-cache-dir -r requirements.txt \
    -i https://pypi.tuna.tsinghua.edu.cn/simple

# 复制应用代码
COPY src/ ./src/
COPY config/ ./config/

# 设置环境变量
ENV PYTHONUNBUFFERED=1
ENV CUDA_VISIBLE_DEVICES=0

# 健康检查
HEALTHCHECK --interval=30s --timeout=10s --start-period=60s --retries=3 \
    CMD python3 src/health_check.py || exit 1

# 启动命令
CMD ["python3", "-m", "uvicorn", "src.main:app", "--host", "0.0.0.0", "--port", "8080"]

version: '3.8'

services:
  # AI Agent 服务
  agent-service:
    build: .
    ports:
      - "8080:8080"
    environment:
      - MODEL_PATH=/models/llama-7b
      - MAX_TOKENS=2048
      - LOG_LEVEL=INFO
    volumes:
      - ./models:/models
      - ./logs:/app/logs
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 30s
      timeout: 10s
      retries: 3

  # Nginx 网关
  nginx:
    image: nginx:1.25-alpine
    ports:
      - "80:80"
      - "443:443"
    volumes:
      - ./nginx/nginx.conf:/etc/nginx/nginx.conf
      - ./nginx/ssl:/etc/nginx/ssl
    depends_on:
      - agent-service

  # Redis 缓存
  redis:
    image: redis:7-alpine
    ports:
      - "6379:6379"
    volumes:
      - redis_data:/data

  # Prometheus 监控
  prometheus:
    image: prom/prometheus:latest
    ports:
      - "9090:9090"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml

volumes:
  redis_data:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: ai-agent
  namespace: ai-platform
spec:
  replicas: 3
  selector:
    matchLabels:
      app: ai-agent
  template:
    metadata:
      labels:
        app: ai-agent
      annotations:
        prometheus.io/scrape: "true"
        prometheus.io/port: "8080"
    spec:
      containers:
      - name: agent
        image: registry.example.com/ai-agent:v1.2.0
        ports:
        - containerPort: 8080
          name: http
        resources:
          requests:
            cpu: "2"
            memory: "4Gi"
            nvidia.com/gpu: "1"
          limits:
            cpu: "4"
            memory: "8Gi"
            nvidia.com/gpu: "1"
        env:
        - name: MODEL_PATH
          value: "/models/llama-7b"
        - name: REDIS_URL
          value: "redis://redis:6379"
        volumeMounts:
        - name: model-storage
          mountPath: /models
        - name: config
          mountPath: /app/config
        livenessProbe:
          httpGet:
            path: /health
            port: 8080
          initialDelaySeconds: 60
          periodSeconds: 30
        readinessProbe:
          httpGet:
            path: /ready
            port: 8080
          initialDelaySeconds: 30
          periodSeconds: 10
      volumes:
      - name: model-storage
        persistentVolumeClaim:
          claimName: model-pvc
      - name: config
        configMap:
          name: agent-config
      nodeSelector:
        gpu-type: nvidia-a100
      tolerations:
      - key: "gpu"
        operator: "Equal"
        value: "true"
        effect: "NoSchedule"

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: ai-agent-hpa
  namespace: ai-platform
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: ai-agent
  minReplicas: 2
  maxReplicas: 20
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 80
  - type: Pods
    pods:
      metric:
        name: requests_per_second
      target:
        type: AverageValue
        averageValue: 100
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
      - type: Percent
        value: 50
        periodSeconds: 60
    scaleUp:
      stabilizationWindowSeconds: 0
      policies:
      - type: Percent
        value: 100
        periodSeconds: 15
      - type: Pods
        value: 4
        periodSeconds: 15
      selectPolicy: Max

apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: ai-agent-ingress
  namespace: ai-platform
  annotations:
    nginx.ingress.kubernetes.io/ssl-redirect: "true"
    nginx.ingress.kubernetes.io/proxy-body-size: "100m"
    nginx.ingress.kubernetes.io/proxy-read-timeout: "300"
    nginx.ingress.kubernetes.io/rate-limit: "100"
    nginx.ingress.kubernetes.io/rate-limit-window: "1m"
spec:
  ingressClassName: nginx
  tls:
  - hosts:
    - ai-agent.example.com
    secretName: ai-tls-secret
  rules:
  - host: ai-agent.example.com
    http:
      paths:
      - path: /api/v1/agent
        pathType: Prefix
        backend:
          service:
            name: ai-agent-service
            port:
              number: 80
      - path: /api/v1/models
        pathType: Prefix
        backend:
          service:
            name: model-service
            port:
              number: 80

#!/bin/bash
# AI Agent 平台自动化部署脚本

# 1. 创建命名空间
kubectl create namespace ai-platform

# 2. 部署 NVIDIA Device Plugin
kubectl apply -f https://raw.githubusercontent.com/NVIDIA/k8s-device-plugin/main/deployments/static/nvidia-device-plugin.yml

# 3. 部署 Nginx Ingress Controller
helm install nginx-ingress ingress-nginx/ingress-nginx \
  --namespace ingress-nginx \
  --create-namespace \
  --set controller.replicaCount=3 \
  --set controller.service.type=LoadBalancer

# 4. 部署 AI Agent 服务
kubectl apply -f k8s/agent-deployment.yaml
kubectl apply -f k8s/agent-service.yaml
kubectl apply -f k8s/agent-hpa.yaml

# 5. 部署模型服务
kubectl apply -f k8s/model-deployment.yaml
kubectl apply -f k8s/model-service.yaml

# 6. 部署监控
helm install prometheus prometheus-community/kube-prometheus-stack \
  --namespace monitoring \
  --create-namespace

# 7. 验证部署
kubectl get pods -n ai-platform
kubectl get svc -n ai-platform
kubectl get ingress -n ai-platform

# 工作进程优化
worker_processes auto;
worker_rlimit_nofile 65535;

# 连接优化
events {
    worker_connections 4096;
    multi_accept on;
    use epoll;
}

# HTTP 优化
http {
    sendfile on;
    tcp_nopush on;
    tcp_nodelay on;
    keepalive_timeout 65;
    keepalive_requests 1000;
    
    # 缓冲区优化
    client_body_buffer_size 10M;
    client_header_buffer_size 1k;
    large_client_header_buffers 4 16k;
    
    # Gzip 压缩
    gzip on;
    gzip_vary on;
    gzip_min_length 1024;
    gzip_comp_level 6;
}

apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
  name: ai-agent-alerts
  namespace: monitoring
spec:
  groups:
  - name: ai-agent.rules
    rules:
    - alert: HighErrorRate
      expr: sum(rate(http_requests_total{status=~"5.."}[5m])) / sum(rate(http_requests_total[5m])) > 0.01
      for: 5m
      labels:
        severity: critical
      annotations:
        summary: AI Agent 错误率过高
        description: 错误率 {{ $value | humanizePercentage }}
    
    - alert: HighLatency
      expr: histogram_quantile(0.99, rate(http_request_duration_seconds_bucket[5m])) > 3
      for: 10m
      labels:
        severity: warning
      annotations:
        summary: AI Agent 延迟过高
        description: P99 延迟 {{ $value }}s
    
    - alert: GPUUtilizationHigh
      expr: DCGM_FI_DEV_GPU_UTIL > 90
      for: 15m
      labels:
        severity: warning
      annotations:
        summary: GPU 利用率过高
        description: GPU 利用率 {{ $value }}%