使用nginx作為前端代理，在reload nginx的時候，發(fā)現(xiàn)nginx會一直處于shutting狀態(tài)，當(dāng)reload nginx次數(shù)多了以后，cup會飆升到100%， gateway的狀態(tài)也會處于busy狀態(tài)。

現(xiàn)象如下：

//nginx進程
$ ps aux | grep nginx
**www      12384  0.6  0.2 110752 37424 ?        SN   Jan20  12:51 nginx: worker process is shutting down**
www      12385  0.1  0.1 102508 29260 ?        SN   Jan20   3:18 nginx: worker process is shutting down
www      12386  0.5  0.2 112744 39616 ?        SN   Jan20  12:45 nginx: worker process is shutting down
www      12387  0.2  0.1 104556 31228 ?        SN   Jan20   5:56 nginx: worker process is shutting down
www      27928  1.0  0.1 102508 28252 ?        SN   11:25   0:08 nginx: worker process
www      27929  0.5  0.1 102508 27932 ?        SN   11:25   0:04 nginx: worker process
www      27930  1.2  0.1 102508 28512 ?        SN   11:25   0:10 nginx: worker process
www      27931  0.2  0.1 102508 27900 ?        SN   11:25   0:02 nginx: worker process
www      29369  0.1  0.1 102508 27712 ?        SN   Jan21   0:52 nginx: worker process is shutting down
www      29370  0.5  0.1 102804 29400 ?        SN   Jan21   3:42 nginx: worker process is shutting down
www      29371  0.2  0.1 102508 28460 ?        SN   Jan21   1:39 nginx: worker process is shutting down
www      29372  0.4  0.1 102804 29360 ?        SN   Jan21   3:23 nginx: worker process is shutting down

再查id為12384的進程發(fā)現(xiàn)，這個進程一直和gateway在連接

$ lsof -i :60877     //60877是從lsof的最后一行取的
COMMAND   PID USER   FD   TYPE     DEVICE SIZE/OFF NODE NAME
php     11789 root   65u  IPv4 2550865341      0t0  TCP mt-web1:8282->mt-web1:60877 (ESTABLISHED)
nginx   12384  www 1184u  IPv4 2550863681      0t0  TCP mt-web1:60877->mt-web1:8282 (ESTABLISHED)

gateway的狀態(tài)如下：發(fā)現(xiàn)會有11789這個進程（當(dāng)nginx reload的次數(shù)多的時候，gateway就會處于busy狀態(tài)。）

Workerman version:3.5.1          PHP version:7.1.6
start time:2017-12-07 09:03:05   run 46 days 4 hours   
load average: 0.14, 0, 0         event-loop:\Workerman\Events\Select
1 workers       4 processes
worker_name     exit_status      exit_count
zhibo-gateway-1 0                0
----------------------------------------------PROCESS STATUS---------------------------------------------------
pid     memory  listening                      worker_name     connections total_request send_fail timers   status
11785   8M      websocket://ip:8282 zhibo-gateway-1 436         32233007      579       3         
11786   8M      websocket://ip:8282 zhibo-gateway-1 431         36117127      644       3         
11788   8M      websocket://ip:8282 zhibo-gateway-1 432         36397854      592       3         
11789   8M      websocket://ip:8282 zhibo-gateway-1 447         33917464      642       3         

因為我們上線新功能后，一般只會reload nginx， 所以懷疑是gateway和nginx鏈接導(dǎo)致的，但是gateway reload后，gateway的進程還是那幾個并沒有重新啟動，這正常嗎？

ps：如果把gateway restart了，gateway的進程id會改變，nginx中的shutting狀態(tài)的進程也會消失，但是gateway restart是不是會斷掉和客戶端的連接呀？