資訊專欄INFORMATION COLUMN
摘要：相比之下，這個(gè)第三方模塊可以對(duì)后端服務(wù)器提供主動(dòng)式的健康狀態(tài)檢測(cè)。向后端發(fā)送的健康檢查包的間隔。如果連續(xù)成功次數(shù)達(dá)到，服務(wù)器就被認(rèn)為是。只有配置塊中支持類型檢查發(fā)送請(qǐng)求，通過(guò)后端的回復(fù)包的狀態(tài)來(lái)判斷后端是否存活。指定后端服務(wù)器的檢查端口。
ngx-healthcheck-module
Health-checker for Nginx upstream servers (support http upstream && stream upstream)
該模塊可以為Nginx提供主動(dòng)式后端服務(wù)器健康檢查的功能（同時(shí)支持四層和七層后端服務(wù)器的健康檢測(cè)）。
Table of ContentsStatus
Description
Installation
Usage
Synopsis && Directive
healthcheck
check
Bugs and Patches
Author
Copyright and License
See Also
StatusThis nginx module is still under development， you can help improve and it.
這個(gè)項(xiàng)目還在開(kāi)發(fā)中完善中，歡迎貢獻(xiàn)代碼，或報(bào)告bug。一起使它變得更好。
Description當(dāng)你使用nginx作為負(fù)載均衡器時(shí)，nginx原生只提供了基本的重試方式來(lái)保證訪問(wèn)到正常的后端服務(wù)器。
相比之下，這個(gè)nginx第三方模塊可以對(duì)后端服務(wù)器提供主動(dòng)式的健康狀態(tài)檢測(cè)。
它維護(hù)了一個(gè)后端服務(wù)器列表，保證新的請(qǐng)求直接發(fā)送到一個(gè)健康的后端服務(wù)器。
主要特性：
同時(shí)支持四層和七層后端服務(wù)器的健康檢測(cè)
四層支持的檢測(cè)類型：tcp / udp / http
七層支持的檢測(cè)類型：http / fastcgi
提供一個(gè)統(tǒng)一的http狀態(tài)查詢接口，輸出格式：html / json / csv
Installationgit clone https://github.com/nginx/nginx/nginx.git git clone https://github.com/zhouchangxun/ngx_healthcheck_module.git cd nginx/; git apply ../ngx_healthcheck_module/nginx-stable-1.12+.patch ./auto/configure --with-stream --add-module=../ngx_healthcheck_module/ make && make install
Back to TOC
Usage nginx.conf exampleuser root;
worker_processes 1;
error_log logs/error.log info;
#pid logs/nginx.pid;
events {
worker_connections 1024;
}
http {
server {
listen 80;
# status interface
location /status {
healthcheck_status;
}
# http front
location / {
proxy_pass http://http-cluster;
}
}
# as a backend server.
server {
listen 8080;
location / {
root html;
}
}
upstream http-cluster {
# simple round-robin
server 127.0.0.1:8080;
server 127.0.0.2:81;
check interval=3000 rise=2 fall=5 timeout=5000 type=http;
check_http_send "GET / HTTP/1.0
";
check_http_expect_alive http_2xx http_3xx;
}
}
stream {
upstream tcp-cluster {
# simple round-robin
server 127.0.0.1:22;
server 192.168.0.2:22;
check interval=3000 rise=2 fall=5 timeout=5000 default_down=true type=tcp;
}
server {
listen 522;
proxy_pass tcp-cluster;
}
upstream udp-cluster {
# simple round-robin
server 127.0.0.1:53;
server 8.8.8.8:53;
check interval=3000 rise=2 fall=5 timeout=5000 default_down=true type=udp;
}
server {
listen 53;
proxy_pass udp-cluster;
}
}
status interface
One typical output is
root@changxun-PC:~/nginx-dev/ngx_healthcheck_module# curl localhost/status
{"servers": {
"total": 6,
"generation": 3,
"http": [
{"index": 0, "upstream": "http-cluster", "name": "127.0.0.1:8080", "status": "up", "rise": 119, "fall": 0, "type": "http", "port": 0},
{"index": 1, "upstream": "http-cluster", "name": "127.0.0.2:81", "status": "down", "rise": 0, "fall": 120, "type": "http", "port": 0}
],
"stream": [
{"index": 0, "upstream": "tcp-cluster", "name": "127.0.0.1:22", "status": "up", "rise": 22, "fall": 0, "type": "tcp", "port": 0},
{"index": 1, "upstream": "tcp-cluster", "name": "192.168.0.2:22", "status": "down", "rise": 0, "fall": 7, "type": "tcp", "port": 0},
{"index": 2, "upstream": "udp-cluster", "name": "127.0.0.1:53", "status": "down", "rise": 0, "fall": 120, "type": "udp", "port": 0},
{"index": 3, "upstream": "udp-cluster", "name": "8.8.8.8:53", "status": "up", "rise": 3, "fall": 0, "type": "udp", "port": 0}
]
}}
root@changxun-PC:~/nginx-dev/ngx_healthcheck_module#
Back to TOC
Synopsis checkSyntax:
check interval=milliseconds
[fall=count] [rise=count] [timeout=milliseconds]
[default_down=true|false] [type=tcp|udp|http] [port=check_port]
Default: interval=30000 fall=5 rise=2 timeout=1000 default_down=true type=tcp
Context: http/upstream || stream/upstream
該指令可以打開(kāi)后端服務(wù)器的健康檢查功能。
Detail:
interval：向后端發(fā)送的健康檢查包的間隔。
fall(fall_count): 如果連續(xù)失敗次數(shù)達(dá)到fall_count，服務(wù)器就被認(rèn)為是down。
rise(rise_count): 如果連續(xù)成功次數(shù)達(dá)到rise_count，服務(wù)器就被認(rèn)為是up。
timeout: 后端健康請(qǐng)求的超時(shí)時(shí)間。
default_down: 設(shè)定初始時(shí)服務(wù)器的狀態(tài)，如果是true，就說(shuō)明默認(rèn)是down的，如果是false，就是up的。
默認(rèn)值是true，也就是一開(kāi)始服務(wù)器認(rèn)為是不可用，要等健康檢查包達(dá)到一定成功次數(shù)以后才會(huì)被認(rèn)為是健康的。
type：健康檢查包的類型，現(xiàn)在支持以下多種類型
tcp：簡(jiǎn)單的tcp連接，如果連接成功，就說(shuō)明后端正常。
udp：簡(jiǎn)單的發(fā)送udp報(bào)文，如果收到icmp error(主機(jī)或端口不可達(dá))，就說(shuō)明后端異常。(只有stream配置塊中支持udp類型檢查)
http：發(fā)送HTTP請(qǐng)求，通過(guò)后端的回復(fù)包的狀態(tài)來(lái)判斷后端是否存活。
port: 指定后端服務(wù)器的檢查端口。你可以指定不同于真實(shí)服務(wù)的后端服務(wù)器的端口，
比如后端提供的是443端口的應(yīng)用，你可以去檢查80端口的狀態(tài)來(lái)判斷后端健康狀況。默認(rèn)是0，表示跟后端server提供真實(shí)服務(wù)的端口一樣。
A example as followed:
stream {
upstream tcp-cluster {
# simple round-robin
server 127.0.0.1:22;
server 192.168.0.2:22;
check interval=3000 rise=2 fall=5 timeout=5000 default_down=true type=tcp;
}
server {
listen 522;
proxy_pass tcp-cluster;
}
...
}
healthcheck
Syntax: healthcheck_status [html|csv|json]
Default: healthcheck_status html
Context: http/server/location
A example as followed:
http {
server {
listen 80;
# status interface
location /status {
healthcheck_status;
}
...
}
Back to TOC
Todo List增加測(cè)試用例
整理、優(yōu)化代碼
規(guī)范代碼中的log輸出
Back to TOC
Bugs and PatchesPlease report bugs
create GitHub Issue,
or submit patches by
new Pull request
Back to TOC
AuthorChance Chou (周長(zhǎng)勛)
Back to TOC
Copyright and LicenseThe health check part is based on Yaoweibin"s
