Nginx中动态upstream域名解析失效问题解决的几种方法

# ❌ 错误示例：仅在启动时解析一次
upstream backend {
    server api.example.com:8080;  # 启动时解析，之后永远不变
}
location / {
    proxy_pass http://backend;
}

http {
    # 1. 配置 DNS 解析器（必须放在 http/server/location 层级）
    resolver 8.8.8.8 8.8.4.4 valid=10s;  # Google DNS，TTL 10秒
    resolver_timeout 5s;
    server {
        listen 80;
        location /api/ {
            # 2. 使用变量强制动态解析
            set $backend "api.example.com";
            proxy_pass http://$backend:8080;
            # 3. 关键：配合 resolver 实现每次请求都解析（或按 TTL 缓存）
            proxy_http_version 1.1;
            proxy_set_header Host $host;
        }
    }
}

http {
    # 针对 Consul 或 Kubernetes CoreDNS 优化
    resolver 127.0.0.1:8600 valid=5s;  # Consul DNS 端口
    resolver_timeout 2s;
    # 可选：指定 DNS 搜索域
    resolver 10.96.0.10 valid=5s ipv6=off;  # K8s CoreDNS，禁用 IPv6 避免延迟
    server {
        location / {
            set $svc "user-service.service.consul";
            proxy_pass http://$svc:8080;
            # 连接池优化：避免频繁 DNS 查询影响性能
            proxy_connect_timeout 2s;
            proxy_send_timeout 5s;
            proxy_read_timeout 10s;
        }
    }
}

http {
    resolver 10.0.0.2 valid=10s;
    # 定义 upstream 但使用域名变量
    upstream dynamic_backend {
        server 0.0.0.0;  # 占位，实际在 location 中解析
        # 健康检查（需要 nginx_upstream_check_module 或商业版）
        check interval=3000 rise=2 fall=3 timeout=1000 type=http;
        check_http_send "HEAD /health HTTP/1.0\r\n\r\n";
        check_http_expect_alive http_2xx http_3xx;
    }
    server {
        location / {
            # 关键：使用变量触发 resolver
            set $backend_nodes "app.service.consul";
            # 通过 Lua 或 split_clients 实现动态选择（见方案三）
            proxy_pass http://$backend_nodes;
        }
    }
}

http {
    resolver 127.0.0.1:8600;
    server {
        location / {
            access_by_lua_block {
                local resolver = require "resty.dns.resolver"
                -- 1. 创建 DNS 客户端
                local dns, err = resolver:new{
                    nameservers = {"127.0.0.1", {"127.0.0.1", 8600}},
                    retrans = 2,
                    timeout = 1000,
                }
                -- 2. 查询 SRV 记录（Consul 推荐）
                local answers, err = dns:query("user-service.service.consul", 
                                               {qtype = dns.TYPE_SRV})
                -- 3. 动态选择节点（加权随机）
                if answers then
                    local target = answers[math.random(1, #answers)]
                    ngx.var.target_host = target.target
                    ngx.var.target_port = target.port
                else
                    -- 降级到静态配置
                    ngx.var.target_host = "127.0.0.1"
                    ngx.var.target_port = "8080"
                end
            }
            set $target_host "";
            set $target_port "";
            proxy_pass http://$target_host:$target_port;
        }
    }
}

upstream backend {
    zone upstream_backend 64k;
    # 关键：resolve 参数启用动态 DNS
    server api.example.com:8080 resolve;
    # 健康检查
    health_check interval=5s fails=3 passes=2;
}
server {
    location / {
        proxy_pass http://backend;
        health_check;
    }
}

# ❌ 错误：直接写死域名，不会触发 resolver
proxy_pass http://api.example.com:8080;
# ✅ 正确：使用变量强制动态解析
set $backend "api.example.com";
proxy_pass http://$backend:8080;

# 错误原因：location 中使用了变量，但上级未定义 resolver
# 解决：确保 resolver 在 http 或 server 层级定义
http {
    resolver 8.8.8.8;  # 必须在这里定义，不能只在 location
}

# 优化：本地 DNS 缓存 + 连接池
resolver 127.0.0.1 valid=5s;  # 指向本地 dnsmasq/unbound
upstream backend {
    keepalive 100;  # 连接池，避免每次新建连接
    keepalive_timeout 60s;
    keepalive_requests 1000;
}

http {
    # DNS 配置（多可用区冗余）
    resolver 10.0.0.2 10.0.0.3 valid=10s ipv6=off;
    resolver_timeout 3s;
    # 日志记录解析详情（调试用）
    log_format dns_log '$remote_addr - $upstream_addr [$time_local] '
                      '"$request" $status $body_bytes_sent '
                      'upstream: "$upstream_addr" '
                      'resolve: "$upstream_http_host"';
    upstream app_servers {
        zone app_zone 64k;
        keepalive 100;
        # 占位 server，实际在 location 中动态解析
        server 127.0.0.1:1 backup;  # 备用，防止启动失败
    }
    server {
        listen 80;
        access_log /var/log/nginx/access.log dns_log;
        location / {
            # 动态解析核心配置
            set $backend_host "app.service.consul";
            set $backend_port "8080";
            proxy_pass http://$backend_host:$backend_port;
            # 错误降级
            proxy_intercept_errors on;
            error_page 502 503 504 = @fallback;
            # 连接优化
            proxy_connect_timeout 3s;
            proxy_next_upstream error timeout http_502 http_503 http_504;
            proxy_next_upstream_tries 2;
        }
        location @fallback {
            internal;
            return 503 '{"error":"Service Unavailable","retry_after":10}';
        }
    }
}