其实反向代理模式很简单,Nginx监听在80端口,做为Web服务端口,而Tornado的Web服务进程监听在808*的内部端口(可以启动多个进程),使用supervisor对Nginx、Tornado服务进程进行统一的管理。
首先看supervisor的配置:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
# supervisor自己的配置及日志切割等 [supervisord] logfile = /opt/logs/supervisord .log logfile_maxbytes = 200MB logfile_backups=10 loglevel = warn pidfile = /opt/logs/supervisord .pid nodaemon = false minfds = 1024 minprocs = 200 umask = 022 identifier = supervisor directory = %(here)s nocleanup = true strip_ansi = false [unix_http_server] file = /opt/logs/supervisord .sock [supervisorctl] serverurl = unix: ///opt/logs/supervisord .sock [rpcinterface:supervisor] supervisor.rpcinterface_factory = supervisor.rpcinterface:make_main_rpcinterface # 服务进程的启动及多端口 [program:MyWeb] command = /opt/bin/python %(here)s /server .py --port=808%(process_num)01d process_name = 808%(process_num)01d numprocs = 4 numprocs_start = 1 autostart = true autorestart = true redirect_stderr = true stdout_logfile = /opt/logs/stdout .log stderr_logfile = /opt/logs/stdout .log # Nginx的管理配置 [program:Nginx] command = /opt/sbin/nginx -c %(here)s /nginx .conf process_name = Nginx numprocs = 1 autostart = true autorestart = true redirect_stderr = true stdout_logfile = /opt/logs/nginx_stdout .log stderr_logfile = /opt/logs/nginx_stdout .log |
启动脚本(可以放到start.sh中):
1
|
/opt/bin/supervisord /opt/conf/supervisor .conf |
重启脚本(可以放到restart.sh中)
1
2
3
4
5
6
7
8
|
#逐个启动MyWeb每个端口进程,不中断服务 for i in "8081 8082 8083 8084" : do /opt/bin/supervisorctl /opt/conf/supervisor .conf restart MyWeb:$i; done #重新加载nginx的配置 /opt/sbin/nginx /opt/conf/nginx .conf -s reload; |
Nginx的部分配置(启动4个服务进程,监听在80端口,并反向代理负载到Tornado的808*端口上):
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
|
worker_processes 4; daemon off; #nginx不能以daemon模式启动 user nobody; http { upstream myweb { server 127.0.0.1:8081; server 127.0.0.1:8082; server 127.0.0.1:8083; server 127.0.0.1:8084; } server { listen 80; server_name localhost; location / { proxy_set_header Host $http_host; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $remote_addr; proxy_set_header X-Scheme $scheme; proxy_read_timeout 300s; proxy_pass http: //myweb ; } } } |
现在Nginx已经反向代理到Tornado的服务进程监听的端口了,那么MyWeb的服务进程如何构建、并如何优雅的重启呢,略过其他代码,介绍一下主进程采用信号停止服务,并重新启动的方法。主进程的启动参数会指定此进程监听的端口,这样supervisor检测到服务进程结束后,会自动启动对应的服务进程。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
|
import signal import tornado.ioloop import tornado.httpserver http_server = None def sig_handler(sig, frame): """信号处理函数 """ tornado.ioloop.IOLoop.instance().add_callback(shutdown) def shutdown(): """进程关闭处理 """ # 停止接受Client连接 global http_server http_server.stop() io_loop = tornado.ioloop.IOLoop.instance() deadline = time.time() + 10 #设置最长强制结束时间 def stop_loop(): now = time.time() if now < deadline: io_loop.add_timeout(now + 1 , stop_loop) else : io_loop.stop() stop_loop() if __name__ = = ‘__main__‘ : # 等待supervisor发送进程结束信号 signal.signal(signal.SIGTERM, sig_handler) signal.signal(signal.SIGINT, sig_handler) app = Application() http_server = tornado.httpserver.HTTPServer(app, xheaders = True ) http_server.listen(tornado.options.options.port) tornado.ioloop.IOLoop.instance().start() |