Nginx 反向代理 Upstream 失败重试和封禁机制

# ...

http {
    upstream demo {
        server  127.0.0.1:8001;  # 默认: max_fails=1 fail_timeout=10s
        server  127.0.0.1:8002;  # 默认: max_fails=1 fail_timeout=10s
        server  127.0.0.1:8003;  # 默认: max_fails=1 fail_timeout=10s
    }
    server {
        listen       8000;
        server_name  localhost;

        location /500 {
            proxy_next_upstream error timeout http_500; # 哪些错误认为是不成功的尝试（unsuccessful attempts），默认为 error timeout，这里的只 http_500 是个例子，强烈不建议设置，否则可能导致异常的节点封禁。
            proxy_next_upstream_tries 2;                # 尝试下一个 server 的次数，默认为 0 不限制
            proxy_pass http://demo; 
        }
    }
    # ...
}

client ---> nginx (:8000) ---> server 1 (:8001)
                           |
                           +-> server 2 (:8002)
                           |
                            -> server 3 (:8003)

package main

import (
	"fmt"
	"net/http"
	"os"
	"sync/atomic"
)

var counter int32

func resp502(w http.ResponseWriter, req *http.Request) {
	atomic.AddInt32(&counter, 1)
	clientcounter := req.URL.Query().Get("counter")
	fmt.Printf("(%s) server counter: %d, client counter:%s\n", os.Args[1], counter, clientcounter)
	w.WriteHeader(500)
	fmt.Fprintf(w, "(%s) server counter: %d", os.Args[1], counter)
}

func main() {
	http.HandleFunc("/500", resp502)
	http.ListenAndServe(fmt.Sprintf(":%s", os.Args[1]), nil)
}

package main

import (
	"fmt"
	"io/ioutil"
	"net/http"
	"os"
	"sync/atomic"
	"time"
)

var counter int32

func req500Path() {
	atomic.AddInt32(&counter, 1)
	resp, err := http.Get(fmt.Sprintf("http://localhost:%s/500?counter=%d", os.Args[1], counter))
	if err != nil {
		fmt.Printf("client counter: %d, error: %s\n", counter, err)
		return
	}
	contentBytes, _ := ioutil.ReadAll(resp.Body)
	fmt.Printf("client counter: %d, status: %d, body: %s\n", counter, resp.StatusCode, string(contentBytes))
}

func main() {
	for i := 0; i < 20; i++ {
		req500Path()
		time.Sleep(1 * time.Second)
	}
}

user  nginx;
worker_processes  auto;

error_log  /var/log/nginx/error.log notice;
pid        /var/run/nginx.pid;


events {
    worker_connections  1024;
}


http {
    include       /etc/nginx/mime.types;
    default_type  application/octet-stream;

    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;

    sendfile        on;
    #tcp_nopush     on;

    keepalive_timeout  65;

    #gzip  on;

    upstream demo {
        server  127.0.0.1:8001;  # 默认: max_fails=1 fail_timeout=10s
        server  127.0.0.1:8002;  # 默认: max_fails=1 fail_timeout=10s
        server  127.0.0.1:8003;  # 默认: max_fails=1 fail_timeout=10s
    }
    server {
        listen       8000;
        server_name  localhost;

        location /500 {
            proxy_next_upstream error timeout http_500; # 哪些错误认为是不成功的尝试（unsuccessful attempts），默认为 error timeout，这里的 http_500 是个例子，强烈不建议设置，否则可能导致异常的节点封禁。
            proxy_next_upstream_tries 2;                # 尝试下一个 server 的次数，默认为 0 不限制
            proxy_pass http://demo; 
        }
    }
    # include /etc/nginx/conf.d/*.conf;
}

# 修改 proxy_next_upstream_tries 重复实验

# 测试
go run ./server 8001  # 终端 1
go run ./server 8002  # 终端 2
go run ./server 8003  # 终端 3
docker run --network host --name demo-nginx -v $(pwd)/nginx.conf:/etc/nginx/nginx.conf:ro nginx:1.23.2  # 终端 4
go run ./client 8000  # 终端 5

# 恢复现场
# ctrl + c 终端 1~5
docker rm -f demo-nginx