Kubernetes / Nginx 金丝雀与灰度发布实战

Posted recently Updated recently

By Administrator

56~72 min read

Kubernetes / Nginx 金丝雀与灰度发布实战

面试指北 · 通用

📊 三种发布策略对比

发布策略	流量分配	风险等级	回滚速度	资源占用	适用场景
蓝绿发布	100% 切换	中等	秒级	2倍资源	关键业务系统
金丝雀发布	5%→20%→50%→100%	低	分钟级	1.2倍资源	高可用要求
灰度发布	按用户群体	低	分钟级	1.5倍资源	A/B测试场景

🐳 Docker + Nginx 实现方案

1. 蓝绿发布

Docker Compose 配置

# docker-compose-blue-green.yml
version: '3.8'

services:
  # 蓝环境 (当前生产)
  gis-app-blue:
    image: gis-backend:v1.0
    container_name: gis-blue
    ports:
      - "8081:8080"
    environment:
      - SPRING_PROFILES_ACTIVE=production
      - DB_HOST=postgres-blue
    networks:
      - gis-network
    labels:
      - "version=blue"
      - "env=production"

  # 绿环境 (新版本)
  gis-app-green:
    image: gis-backend:v2.0
    container_name: gis-green
    ports:
      - "8082:8080"
    environment:
      - SPRING_PROFILES_ACTIVE=production
      - DB_HOST=postgres-green
    networks:
      - gis-network
    labels:
      - "version=green"
      - "env=staging"

  # Nginx 负载均衡器
  nginx:
    image: nginx:stable-alpine
    container_name: nginx-lb
    ports:
      - "80:80"
      - "443:443"
    volumes:
      - ./nginx/nginx.conf:/etc/nginx/nginx.conf
      - ./nginx/conf.d:/etc/nginx/conf.d
    networks:
      - gis-network
    depends_on:
      - gis-app-blue
      - gis-app-green

networks:
  gis-network:
    driver: bridge

Nginx 蓝绿切换配置

# nginx/conf.d/blue-green.conf
upstream backend {
    # 当前指向蓝环境
    server gis-app-blue:8080 weight=100;
    # 绿环境准备就绪但权重为0
    server gis-app-green:8080 weight=0 backup;
}

server {
    listen 80;
    server_name gis.example.com;

    # 健康检查端点
    location /health {
        access_log off;
        return 200 "healthy\n";
        add_header Content-Type text/plain;
    }

    # API流量代理
    location /api/ {
        proxy_pass http://backend;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;

        # 超时配置
        proxy_connect_timeout 5s;
        proxy_send_timeout 10s;
        proxy_read_timeout 10s;
    }

    # 静态资源
    location / {
        proxy_pass http://backend;
        proxy_set_header Host $host;
    }
}

# 管理端口 - 用于动态切换
server {
    listen 8080;
    server_name localhost;

    # 切换到绿环境
    location /switch-to-green {
        access_log off;
        content_by_lua_block {
            -- 这里需要安装 lua 模块
            -- 动态修改 upstream 配置
            ngx.say("Switching to green environment...")
        }
    }

    # 切换到蓝环境  
    location /switch-to-blue {
        access_log off;
        content_by_lua_block {
            ngx.say("Switching to blue environment...")
        }
    }
}

蓝绿切换脚本

#!/bin/bash
# blue-green-deploy.sh

set -e

CURRENT_ENV=""
NEW_ENV=""
NGINX_CONTAINER="nginx-lb"

# 检查当前环境
check_current_env() {
    CURRENT=$(docker exec $NGINX_CONTAINER nginx -T 2>/dev/null | grep "server gis-app-" | grep "weight=100" | head -1)
    if echo "$CURRENT" | grep -q "blue"; then
        CURRENT_ENV="blue"
        NEW_ENV="green"
    else
        CURRENT_ENV="green"
        NEW_ENV="blue"
    fi
    echo "当前环境: $CURRENT_ENV, 目标环境: $NEW_ENV"
}

# 健康检查
health_check() {
    local env=$1
    local port=""

    if [ "$env" = "blue" ]; then
        port="8081"
    else
        port="8082"
    fi

    echo "正在检查 $env 环境健康状态..."
    for i in {1..30}; do
        if curl -sf "http://localhost:$port/actuator/health" > /dev/null; then
            echo "$env 环境健康检查通过"
            return 0
        fi
        echo "等待 $env 环境启动... ($i/30)"
        sleep 10
    done

    echo "$env 环境健康检查失败"
    return 1
}

# 执行切换
switch_environment() {
    echo "开始切换到 $NEW_ENV 环境..."

    # 生成新的 nginx 配置
    cat > ./nginx/conf.d/blue-green.conf << EOF
upstream backend {
    server gis-app-$NEW_ENV:8080 weight=100;
    server gis-app-$CURRENT_ENV:8080 weight=0 backup;
}
# ... 其余配置保持不变
EOF

    # 重新加载 nginx 配置
    docker exec $NGINX_CONTAINER nginx -s reload

    echo "切换完成！当前环境: $NEW_ENV"
}

# 回滚操作
rollback() {
    echo "正在回滚到 $CURRENT_ENV 环境..."

    cat > ./nginx/conf.d/blue-green.conf << EOF
upstream backend {
    server gis-app-$CURRENT_ENV:8080 weight=100;
    server gis-app-$NEW_ENV:8080 weight=0 backup;
}
EOF

    docker exec $NGINX_CONTAINER nginx -s reload
    echo "回滚完成！"
}

# 主流程
main() {
    check_current_env

    # 启动新环境
    docker-compose -f docker-compose-blue-green.yml up -d gis-app-$NEW_ENV

    # 健康检查
    if health_check $NEW_ENV; then
        switch_environment

        # 验证切换后的状态
        sleep 5
        if health_check $NEW_ENV; then
            echo "部署成功！"

            # 停止旧环境 (可选)
            # docker-compose -f docker-compose-blue-green.yml stop gis-app-$CURRENT_ENV
        else
            echo "切换后健康检查失败，正在回滚..."
            rollback
            exit 1
        fi
    else
        echo "新环境启动失败"
        exit 1
    fi
}

# 检查命令行参数
case "${1:-deploy}" in
    deploy)
        main
        ;;
    rollback)
        check_current_env
        rollback
        ;;
    status)
        check_current_env
        ;;
    *)
        echo "Usage: $0 {deploy|rollback|status}"
        exit 1
        ;;
esac

2. 金丝雀发布

Nginx 金丝雀配置

# nginx/conf.d/canary.conf
# 稳定版本
upstream stable {
    server gis-app-stable:8080;
}

# 金丝雀版本
upstream canary {
    server gis-app-canary:8080;
}

# 基于权重的金丝雀发布
upstream backend_weighted {
    server gis-app-stable:8080 weight=95;  # 95% 流量
    server gis-app-canary:8080 weight=5;   # 5% 流量
}

# 基于请求头的金丝雀发布
map $http_x_canary $pool {
    default "stable";
    "true" "canary";
}

# 基于 Cookie 的金丝雀发布
map $cookie_canary $canary_pool {
    default "stable";
    "enabled" "canary";
}

server {
    listen 80;
    server_name gis.example.com;

    # 设置变量用于路由决策
    set $upstream_pool "";

    # 路由逻辑
    access_by_lua_block {
        local canary_header = ngx.var.http_x_canary
        local canary_cookie = ngx.var.cookie_canary
        local remote_addr = ngx.var.remote_addr

        -- 优先级：请求头 > Cookie > IP 哈希
        if canary_header == "true" then
            ngx.var.upstream_pool = "canary"
        elseif canary_cookie == "enabled" then
            ngx.var.upstream_pool = "canary"
        else
            -- 基于 IP 的一致性哈希，5% 流量到金丝雀
            local hash = ngx.crc32_long(remote_addr)
            if hash % 100 < 5 then
                ngx.var.upstream_pool = "canary"
            else
                ngx.var.upstream_pool = "stable"
            end
        end
    }

    location /api/ {
        # 根据变量选择后端
        proxy_pass http://$upstream_pool;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;

        # 添加响应头标识版本
        add_header X-Served-By $upstream_pool;
    }
}

# 管理接口
server {
    listen 8080;
    server_name localhost;

    # 调整金丝雀流量比例
    location ~ ^/canary/weight/(\d+)$ {
        content_by_lua_block {
            local weight = tonumber(ngx.var[1])
            if weight and weight >= 0 and weight <= 100 then
                -- 动态调整权重配置
                -- 这里需要实现动态更新 upstream 的逻辑
                ngx.say("Canary weight set to: " .. weight .. "%")
            else
                ngx.status = 400
                ngx.say("Invalid weight. Must be 0-100")
            end
        }
    }
}

金丝雀发布自动化脚本

#!/bin/bash
# canary-deploy.sh

STABLE_IMAGE="gis-backend:v1.0"
CANARY_IMAGE="gis-backend:v2.0"
CANARY_WEIGHTS=(5 20 50 100)  # 发布阶段的流量比例
STAGE_DURATION=300  # 每个阶段持续时间（秒）

deploy_canary() {
    echo "开始金丝雀部署..."

    # 启动金丝雀版本
    docker run -d \
        --name gis-app-canary \
        --network gis-network \
        -p 8083:8080 \
        $CANARY_IMAGE

    # 等待启动
    sleep 30

    # 健康检查
    if ! curl -sf "http://localhost:8083/actuator/health"; then
        echo "金丝雀版本启动失败"
        cleanup_canary
        exit 1
    fi

    echo "金丝雀版本启动成功"
}

update_weight() {
    local weight=$1
    echo "调整金丝雀流量到 $weight%"

    # 更新 nginx 配置
    sed -i "s/server gis-app-canary:8080 weight=[0-9]*/server gis-app-canary:8080 weight=$weight/" \
        ./nginx/conf.d/canary.conf

    # 重新加载配置
    docker exec nginx-lb nginx -s reload
}

monitor_metrics() {
    local duration=$1
    echo "监控指标 $duration 秒..."

    local start_time=$(date +%s)
    local end_time=$((start_time + duration))

    while [ $(date +%s) -lt $end_time ]; do
        # 检查错误率
        local error_rate=$(curl -s "http://localhost:9090/api/v1/query?query=rate(http_requests_total{status=~\"5..\"}[5m])" | jq -r '.data.result[0].value[1] // 0')

        # 检查响应时间
        local avg_response_time=$(curl -s "http://localhost:9090/api/v1/query?query=rate(http_request_duration_seconds_sum[5m])/rate(http_request_duration_seconds_count[5m])" | jq -r '.data.result[0].value[1] // 0')

        echo "错误率: $error_rate, 平均响应时间: ${avg_response_time}s"

        # 如果指标异常，立即回滚
        if (( $(echo "$error_rate > 0.01" | bc -l) )); then
            echo "错误率过高，正在回滚..."
            return 1
        fi

        sleep 30
    done

    return 0
}

cleanup_canary() {
    echo "清理金丝雀环境..."
    docker stop gis-app-canary 2>/dev/null || true
    docker rm gis-app-canary 2>/dev/null || true
}

rollback() {
    echo "正在回滚..."
    update_weight 0
    cleanup_canary
    echo "回滚完成"
}

main() {
    deploy_canary

    for weight in "${CANARY_WEIGHTS[@]}"; do
        update_weight $weight

        if [ $weight -eq 100 ]; then
            echo "金丝雀发布完成，切换到新版本"
            # 停止旧版本
            docker stop gis-app-stable
            docker rename gis-app-canary gis-app-stable
            break
        fi

        if ! monitor_metrics $STAGE_DURATION; then
            rollback
            exit 1
        fi

        echo "阶段 $weight% 成功，继续下一阶段"
    done
}

case "${1:-deploy}" in
    deploy)
        main
        ;;
    rollback)
        rollback
        ;;
    *)
        echo "Usage: $0 {deploy|rollback}"
        exit 1
        ;;
esac

☸️ Kubernetes 实现方案

1. 基于 Kubernetes Deployment 的蓝绿发布

应用部署清单

# k8s/blue-green/gis-app-blue.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: gis-app-blue
  namespace: gis-system
  labels:
    app: gis-app
    version: blue
spec:
  replicas: 3
  selector:
    matchLabels:
      app: gis-app
      version: blue
  template:
    metadata:
      labels:
        app: gis-app
        version: blue
    spec:
      containers:
      - name: gis-app
        image: gis-backend:v1.0
        ports:
        - containerPort: 8080
          name: http
        env:
        - name: SPRING_PROFILES_ACTIVE
          value: "production"
        - name: VERSION
          value: "blue"
        resources:
          requests:
            memory: "1Gi"
            cpu: "500m"
          limits:
            memory: "2Gi"
            cpu: "1000m"
        livenessProbe:
          httpGet:
            path: /actuator/health
            port: 8080
          initialDelaySeconds: 60
          periodSeconds: 10
        readinessProbe:
          httpGet:
            path: /actuator/health/readiness
            port: 8080
          initialDelaySeconds: 30
          periodSeconds: 5

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: gis-app-green
  namespace: gis-system
  labels:
    app: gis-app
    version: green
spec:
  replicas: 0  # 初始为0，部署时扩容
  selector:
    matchLabels:
      app: gis-app
      version: green
  template:
    metadata:
      labels:
        app: gis-app
        version: green
    spec:
      containers:
      - name: gis-app
        image: gis-backend:v2.0
        ports:
        - containerPort: 8080
          name: http
        env:
        - name: SPRING_PROFILES_ACTIVE
          value: "production"
        - name: VERSION
          value: "green"
        resources:
          requests:
            memory: "1Gi"
            cpu: "500m"
          limits:
            memory: "2Gi"
            cpu: "1000m"
        livenessProbe:
          httpGet:
            path: /actuator/health
            port: 8080
          initialDelaySeconds: 60
          periodSeconds: 10
        readinessProbe:
          httpGet:
            path: /actuator/health/readiness
            port: 8080
          initialDelaySeconds: 30
          periodSeconds: 5

Service 配置

# k8s/blue-green/service.yaml
apiVersion: v1
kind: Service
metadata:
  name: gis-app-service
  namespace: gis-system
spec:
  selector:
    app: gis-app
    version: blue  # 初始指向蓝环境
  ports:
  - port: 80
    targetPort: 8080
    protocol: TCP
    name: http
  type: ClusterIP

---
# 用于蓝绿切换的服务
apiVersion: v1
kind: Service
metadata:
  name: gis-app-blue
  namespace: gis-system
spec:
  selector:
    app: gis-app
    version: blue
  ports:
  - port: 80
    targetPort: 8080
    protocol: TCP

---
apiVersion: v1
kind: Service
metadata:
  name: gis-app-green
  namespace: gis-system
spec:
  selector:
    app: gis-app
    version: green
  ports:
  - port: 80
    targetPort: 8080
    protocol: TCP

Nginx Ingress 配置

# k8s/blue-green/ingress.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: gis-app-ingress
  namespace: gis-system
  annotations:
    kubernetes.io/ingress.class: "nginx"
    nginx.ingress.kubernetes.io/rewrite-target: /
    nginx.ingress.kubernetes.io/ssl-redirect: "false"
    # 自定义 Nginx 配置
    nginx.ingress.kubernetes.io/configuration-snippet: |
      more_set_headers "X-Served-By: $service_name";
      more_set_headers "X-Pod-Name: $hostname";
spec:
  rules:
  - host: gis.example.com
    http:
      paths:
      - path: /
        pathType: Prefix
        backend:
          service:
            name: gis-app-service
            port:
              number: 80

---
# 管理入口，用于蓝绿切换
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: gis-app-admin
  namespace: gis-system
  annotations:
    kubernetes.io/ingress.class: "nginx"
    nginx.ingress.kubernetes.io/rewrite-target: /
spec:
  rules:
  - host: gis-admin.example.com
    http:
      paths:
      - path: /blue
        pathType: Prefix
        backend:
          service:
            name: gis-app-blue
            port:
              number: 80
      - path: /green
        pathType: Prefix
        backend:
          service:
            name: gis-app-green
            port:
              number: 80

蓝绿发布自动化脚本（K8s版）

#!/bin/bash
# k8s-blue-green-deploy.sh

NAMESPACE="gis-system"
APP_NAME="gis-app"
NEW_IMAGE="$1"

if [ -z "$NEW_IMAGE" ]; then
    echo "Usage: $0 <new-image>"
    exit 1
fi

# 获取当前活动环境
get_current_env() {
    local current_selector=$(kubectl get service gis-app-service -n $NAMESPACE -o jsonpath='{.spec.selector.version}')
    echo $current_selector
}

# 获取目标环境
get_target_env() {
    local current=$1
    if [ "$current" = "blue" ]; then
        echo "green"
    else
        echo "blue"
    fi
}

# 部署新版本
deploy_new_version() {
    local target_env=$1
    local image=$2

    echo "正在部署 $target_env 环境，镜像: $image"

    # 更新镜像
    kubectl set image deployment/gis-app-$target_env gis-app=$image -n $NAMESPACE

    # 扩容到3个副本
    kubectl scale deployment gis-app-$target_env --replicas=3 -n $NAMESPACE

    # 等待就绪
    kubectl wait --for=condition=available --timeout=600s deployment/gis-app-$target_env -n $NAMESPACE

    if [ $? -ne 0 ]; then
        echo "$target_env 环境部署失败"
        return 1
    fi

    echo "$target_env 环境部署成功"
    return 0
}

# 健康检查
health_check() {
    local env=$1
    echo "正在进行 $env 环境健康检查..."

    # 通过服务进行健康检查
    for i in {1..30}; do
        if kubectl exec -n $NAMESPACE deployment/gis-app-$env -- curl -sf http://localhost:8080/actuator/health > /dev/null 2>&1; then
            echo "$env 环境健康检查通过"
            return 0
        fi
        echo "等待 $env 环境就绪... ($i/30)"
        sleep 10
    done

    echo "$env 环境健康检查失败"
    return 1
}

# 切换流量
switch_traffic() {
    local target_env=$1
    echo "正在切换流量到 $target_env 环境..."

    kubectl patch service gis-app-service -n $NAMESPACE -p '{"spec":{"selector":{"version":"'$target_env'"}}}'

    if [ $? -eq 0 ]; then
        echo "流量切换成功"
        return 0
    else
        echo "流量切换失败"
        return 1
    fi
}

# 清理旧环境
cleanup_old_env() {
    local old_env=$1
    echo "正在缩容 $old_env 环境..."

    kubectl scale deployment gis-app-$old_env --replicas=0 -n $NAMESPACE
    echo "$old_env 环境已缩容"
}

# 回滚操作
rollback() {
    local original_env=$1
    echo "正在回滚到 $original_env 环境..."

    switch_traffic $original_env
    echo "回滚完成"
}

# 主流程
main() {
    local current_env=$(get_current_env)
    local target_env=$(get_target_env $current_env)

    echo "当前环境: $current_env"
    echo "目标环境: $target_env"
    echo "新镜像: $NEW_IMAGE"

    # 部署新版本
    if ! deploy_new_version $target_env $NEW_IMAGE; then
        echo "部署失败"
        exit 1
    fi

    # 健康检查
    if ! health_check $target_env; then
        echo "健康检查失败，正在清理..."
        kubectl scale deployment gis-app-$target_env --replicas=0 -n $NAMESPACE
        exit 1
    fi

    # 切换流量
    if ! switch_traffic $target_env; then
        echo "流量切换失败，正在回滚..."
        rollback $current_env
        exit 1
    fi

    # 验证切换后状态
    sleep 10
    if health_check $target_env; then
        echo "蓝绿部署成功！"

        # 等待5分钟后清理旧环境
        echo "5分钟后将清理 $current_env 环境..."
        sleep 300
        cleanup_old_env $current_env
    else
        echo "切换后验证失败，正在回滚..."
        rollback $current_env
        exit 1
    fi
}

# 执行主流程
main

2. 基于 Istio 的金丝雀发布

VirtualService 配置

# k8s/canary/virtual-service.yaml
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:
  name: gis-app
  namespace: gis-system
spec:
  hosts:
  - gis.example.com
  http:
  # 基于请求头的路由
  - match:
    - headers:
        x-canary:
          exact: "true"
    route:
    - destination:
        host: gis-app-service
        subset: canary
  # 基于权重的路由
  - route:
    - destination:
        host: gis-app-service
        subset: stable
      weight: 95
    - destination:
        host: gis-app-service
        subset: canary
      weight: 5

---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:
  name: gis-app
  namespace: gis-system
spec:
  host: gis-app-service
  subsets:
  - name: stable
    labels:
      version: stable
  - name: canary
    labels:
      version: canary
  trafficPolicy:
    connectionPool:
      tcp:
        maxConnections: 100
      http:
        http1MaxPendingRequests: 10
        maxRequestsPerConnection: 2
    outlierDetection:
      consecutiveErrors: 3
      interval: 30s
      baseEjectionTime: 30s

Flagger 自动化金丝雀发布

# k8s/canary/flagger-canary.yaml
apiVersion: flagger.app/v1beta1
kind: Canary
metadata:
  name: gis-app
  namespace: gis-system
spec:
  # 目标部署
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: gis-app
  # 服务配置
  service:
    port: 80
    targetPort: 8080
    gateways:
    - gis-gateway
    hosts:
    - gis.example.com
  # 分析配置
  analysis:
    # 分析间隔
    interval: 1m
    # 分析阈值
    threshold: 5
    # 最大权重
    maxWeight: 50
    # 权重增长步长
    stepWeight: 10
    # 成功率阈值
    metrics:
    - name: request-success-rate
      thresholdRange:
        min: 99
      interval: 1m
    - name: request-duration
      thresholdRange:
        max: 500
      interval: 1m
    # Webhook 测试
    webhooks:
    - name: load-test
      url: http://flagger-loadtester.test-system/
      timeout: 5s
      metadata:
        cmd: "hey -z 1m -q 10 -c 2 http://gis.example.com/api/health"

📊 监控和告警配置

Prometheus 监控规则

# monitoring/prometheus-rules.yaml
apiVersion: monitoring.coreos.com/v1
kind: PrometheusRule
metadata:
  name: gis-app-canary
  namespace: gis-system
spec:
  groups:
  - name: gis-app.rules
    rules:
    # 错误率告警
    - alert: HighErrorRate
      expr: |
        (
          rate(http_requests_total{job="gis-app",status=~"5.."}[5m])
          /
          rate(http_requests_total{job="gis-app"}[5m])
        ) > 0.01
      for: 2m
      labels:
        severity: critical
      annotations:
        summary: "GIS应用错误率过高"
        description: "{{ $labels.version }} 版本错误率 {{ $value | humanizePercentage }}"

    # 响应时间告警
    - alert: HighLatency
      expr: |
        histogram_quantile(0.95,
          rate(http_request_duration_seconds_bucket{job="gis-app"}[5m])
        ) > 1
      for: 5m
      labels:
        severity: warning
      annotations:
        summary: "GIS应用响应时间过长"
        description: "{{ $labels.version }} 版本 95% 响应时间超过 1 秒"

Grafana 监控大盘

{
  "dashboard": {
    "title": "GIS Application Canary Deployment",
    "panels": [
      {
        "title": "Request Rate by Version",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(http_requests_total{job=\"gis-app\"}[5m])",
            "legendFormat": "{{version}} - {{status}}"
          }
        ]
      },
      {
        "title": "Error Rate by Version",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(http_requests_total{job=\"gis-app\",status=~\"5..\"}[5m]) / rate(http_requests_total{job=\"gis-app\"}[5m])",
            "legendFormat": "{{version}} Error Rate"
          }
        ]
      },
      {
        "title": "Response Time P95",
        "type": "graph",
        "targets": [
          {
            "expr": "histogram_quantile(0.95, rate(http_request_duration_seconds_bucket{job=\"gis-app\"}[5m]))",
            "legendFormat": "{{version}} P95"
          }
        ]
      }
    ]
  }
}

🔄 CI/CD 集成

GitLab CI 配置

# .gitlab-ci.yml
stages:
  - build
  - test
  - deploy-canary
  - promote

variables:
  DOCKER_REGISTRY: "registry.example.com"
  APP_NAME: "gis-backend"
  KUBE_NAMESPACE: "gis-system"

build:
  stage: build
  script:
    - docker build -t $DOCKER_REGISTRY/$APP_NAME:$CI_COMMIT_SHA .
    - docker push $DOCKER_REGISTRY/$APP_NAME:$CI_COMMIT_SHA

deploy-canary:
  stage: deploy-canary
  script:
    - ./scripts/k8s-canary-deploy.sh $DOCKER_REGISTRY/$APP_NAME:$CI_COMMIT_SHA
  environment:
    name: production-canary
  when: manual

promote-production:
  stage: promote
  script:
    - ./scripts/k8s-promote-canary.sh
  environment:
    name: production
  when: manual
  dependencies:
    - deploy-canary

🎯 最佳实践总结

选择建议

小团队/简单应用: Docker + Nginx
中大型团队: Kubernetes + Ingress Controller
微服务架构: Istio + Flagger
高可用要求: 蓝绿发布
渐进式验证: 金丝雀发布

关键要点

健康检查: 完善的应用健康检查接口
监控告警: 实时监控关键指标
自动回滚: 检测到异常时自动回滚
流量控制: 精确的流量分配和切换
测试策略: 自动化测试验证新版本

这套方案可以根据你们的具体需求进行调整和优化！

面试指北

License: CC BY 4.0