flask 的网站部署了一天， nginx 依然无法与 uwsgi 通讯，系统 Fedora 23。希望搭建过 Flask 环境的大牛看看。

[fyyz]

nginx 的配置（开了 81 端口，方便调试）：

[root@localhost ~]# cat /etc/nginx/conf.d/dev.auth.lan.conf 
server
{
    listen          81;
    server_name     dev.auth.lan;
    root            /var/web/dev.auth.lan;

    location /
    {
        include     uwsgi_params;
        uwsgi_pass  127.0.0.1:8000;
    }
}

uwsgi 的配置：

[root@localhost dev.auth.lan]# cat /etc/uwsgi.ini 
[uwsgi]
uid = uwsgi 
gid = uwsgi
pidfile = /run/uwsgi/uwsgi.pid
emperor = /etc/uwsgi.d
stats = /run/uwsgi/stats.sock
emperor-tyrant = true
cap = setgid,setuid

[root@localhost ~]# cat /etc/uwsgi.d/dev.auth.lan.ini 
[uwsgi]

socket = :8000

chdir = /var/web/dev.auth.lan
wsgi-file = manager.py

master = true
process = 10
vacuum = true

daemonize = /var/log/dev.auth.lan.uwsgi.log

nginx 的运行情况（实心圆圈是绿灯）：

[root@localhost ~]# systemctl status nginx -l
● nginx.service - The nginx HTTP and reverse proxy server
   Loaded: loaded (/usr/lib/systemd/system/nginx.service; enabled; vendor preset: disabled)
   Active: active (running) since 六 2016-01-30 06:03:28 PST; 25min ago
  Process: 936 ExecStart=/usr/sbin/nginx (code=exited, status=0/SUCCESS)
  Process: 861 ExecStartPre=/usr/sbin/nginx -t (code=exited, status=0/SUCCESS)
  Process: 848 ExecStartPre=/usr/bin/rm -f /run/nginx.pid (code=exited, status=0/SUCCESS)
 Main PID: 991 (nginx)
   CGroup: /system.slice/nginx.service
           ├─991 nginx: master process /usr/sbin/ngin
           └─993 nginx: worker proces

1 月 30 06:03:28 localhost.localdomain systemd[1]: Starting The nginx HTTP and reverse proxy server...
1 月 30 06:03:28 localhost.localdomain nginx[861]: nginx: the configuration file /etc/nginx/nginx.conf syntax is ok
1 月 30 06:03:28 localhost.localdomain nginx[861]: nginx: configuration file /etc/nginx/nginx.conf test is successful
1 月 30 06:03:28 localhost.localdomain systemd[1]: nginx.service: Failed to read PID from file /run/nginx.pid: Invalid argument
1 月 30 06:03:28 localhost.localdomain systemd[1]: Started The nginx HTTP and reverse proxy server.

uwsgi 的运行情况（也是绿灯）：

[root@localhost ~]# systemctl status uwsgi -l
● uwsgi.service - uWSGI Emperor Service
   Loaded: loaded (/usr/lib/systemd/system/uwsgi.service; enabled; vendor preset: disabled)
   Active: active (running) since 六 2016-01-30 06:32:29 PST; 2s ago
  Process: 2277 ExecStartPre=/bin/chown uwsgi:uwsgi /run/uwsgi (code=exited, status=0/SUCCESS)
  Process: 2274 ExecStartPre=/bin/mkdir -p /run/uwsgi (code=exited, status=0/SUCCESS)
 Main PID: 2281 (uwsgi)
   Status: "The Emperor is governing 1 vassals"
   CGroup: /system.slice/uwsgi.service
           ├─2281 /usr/sbin/uwsgi --ini /etc/uwsgi.ini
           ├─2283 /usr/sbin/uwsgi --ini /etc/uwsgi.ini
           ├─2284 /usr/sbin/uwsgi --ini dev.auth.lan.ini
           └─2285 /usr/sbin/uwsgi --ini dev.auth.lan.ini

1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: your server socket listen backlog is limited to 100 connections
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: your mercy for graceful operations on workers is 60 seconds
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: mapped 145536 bytes (142 KB) for 1 cores
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: *** Operational MODE: single process ***
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: *** no app loaded. going in full dynamic mode ***
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: *** uWSGI is running in multiple interpreter mode ***
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: spawned uWSGI master process (pid: 2284)
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: Sat Jan 30 06:32:29 2016 - [emperor] vassal dev.auth.lan.ini has been spawned
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: spawned uWSGI worker 1 (pid: 2285, cores: 1)
1 月 30 06:32:29 localhost.localdomain uwsgi[2281]: Sat Jan 30 06:32:29 2016 - [emperor] vassal dev.auth.lan.ini is ready to accept requests

端口也确实开了：

[root@localhost ~]# netstat -lpn | grep 8000
tcp        0      0 0.0.0.0:8000            0.0.0.0:*               LISTEN      2284/uwsgi

项目目录下也确实有 manager.py

[root@localhost dev.auth.lan]# pwd
/var/web/dev.auth.lan

[root@localhost dev.auth.lan]# ls -alh
总用量 4.0K
drwxr-xr-x. 3 root root  39 1 月  30 06:36 .
drwxr-xr-x. 3 root root  26 1 月  30 02:32 ..
-rwxr-xr-x. 1 root root 132 1 月  30 06:10 manager.py
drwxr-xr-x. 5 root root  82 1 月  30 05:02 project

[root@localhost dev.auth.lan]# cat manager.py
#!/usr/bin/env python3

from flask import Flask

app = Flask(__name__)

@app.route('/')
def index():
    return "<span style='color:red'>I am app 1</span>"
[root@localhost dev.auth.lan]#

但是用浏览器访问那台服务器的 81 端口，就是报 502 ， nginx 的日志（太长，加了换行进去）：

2016/01/30 06:37:59 [crit] 993#0: *25 connect() to 127.0.0.1:8000 failed 
(13: Permission denied) 
while connecting to upstream, client: 192.168.124.1, 
server: dev.auth.lan, request: "GET / HTTP/1.1", 
upstream: "uwsgi://127.0.0.1:8000", host: "192.168.124.237:81"

@ethego
直接写 module = manager 依然不行，还是同样的报错，试了很多方法， uwsgi 的报错没了，但是依然是 502 。

[root@localhost dev.auth.lan]# cat /etc/uwsgi.d/dev.auth.lan.ini 
[uwsgi]

socket = :8000

chdir = /var/web/dev.auth.lan
processes = 2
wsgi-file = /var/web/dev.auth.lan/manager.py
callable = app
plugins = python3

master = true
vacuum = true

[root@localhost dev.auth.lan]# systemctl status uwsgi
● uwsgi.service - uWSGI Emperor Service
   Loaded: loaded (/usr/lib/systemd/system/uwsgi.service; enabled; vendor preset: disabled)
   Active: active (running) since 六 2016-01-30 08:24:34 PST; 5s ago
  Process: 4797 ExecStartPre=/bin/chown uwsgi:uwsgi /run/uwsgi (code=exited, status=0/SUCCESS)
  Process: 4793 ExecStartPre=/bin/mkdir -p /run/uwsgi (code=exited, status=0/SUCCESS)
 Main PID: 4801 (uwsgi)
   Status: "The Emperor is governing 1 vassals"
   CGroup: /system.slice/uwsgi.service
           ├─4801 /usr/sbin/uwsgi --ini /etc/uwsgi.ini
           ├─4803 /usr/sbin/uwsgi --ini /etc/uwsgi.ini
           ├─4804 /usr/sbin/uwsgi --ini dev.auth.lan.ini
           ├─4806 /usr/sbin/uwsgi --ini dev.auth.lan.ini
           └─4807 /usr/sbin/uwsgi --ini dev.auth.lan.ini

1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: your mercy for graceful operations on workers is 60 seconds
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: mapped 218304 bytes (213 KB) for 2 cores
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: *** Operational MODE: preforking ***
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: WSGI app 0 (mountpoint='') ready in 1 seconds on interpreter 0x9b93c0 pid: 4804 (default app)
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: *** uWSGI is running in multiple interpreter mode ***
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: spawned uWSGI master process (pid: 4804)
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: Sat Jan 30 08:24:35 2016 - [emperor] vassal dev.auth.lan.ini has been spawned
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: spawned uWSGI worker 1 (pid: 4806, cores: 1)
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: spawned uWSGI worker 2 (pid: 4807, cores: 1)
1 月 30 08:24:35 localhost.localdomain uwsgi[4801]: Sat Jan 30 08:24:35 2016 - [emperor] vassal dev.auth.lan.ini is ready to accept requests

Fedora 上还有一个问题，就是默认是开启 SELINUX 的，按照上述步骤做好以后，再关闭 SELINUX 即可让网站运行起来了。

非常感谢大家的解答。

[ethego]

*** no app loaded. going in full dynamic mode ***
看到这玩意没有？

[ethego]

wsgi 容器里没装东西

[fyyz]

@ethego
非常感谢你的回复。
不过，我不是很清楚，你说的这个 wsgi 容器是什么玩意？依照我上面的配置，我应该如何更改呢？

[ethego]

wsgi-file = manager.py 换成 module = manager 或者 module = manager:app

[fyyz]

直接回复似乎不支持 markdown ，我写 APPEND 里了。
@ethego

[billlee]

看到有 connect() to 127.0.0.1:8000 failed (13: Permission denied)
也许是 SELinux 的问题？

[squid157]

@billlee
我也这么猜。我因为不会配置，都是直接关闭 SELinux

[fyyz]

非常感谢 @billlee @squid157 这两位，当然还有楼上的 @ethego 。目前网站已经成功跑起来了。

[Ansen]

uwsgi 你用 http ，不要用 sockt
sockt 不是这样配置的（ nginx ）

[jings]

mark

[aabbccli]

用 SOCK 文件的话， NGINX 应该是这样

upstream first {
server unix:///var/www/site/site.sock;
}

server {
listen 80;
server_name localhost;


location / {

include /etc/nginx/uwsgi_params;
uwsgi_pass first;

}

}

[fyyz]

@Ansen 为什么不要用 socket ？

[Ansen]

@fyyz 你 nginx 配的是 http ， uwsgi 用的是 sockt