因为我的个人网站 restran.net 已经启用,博客园的内容已经不再更新。请访问我的个人网站获取这篇文章的最新内容,CentOS 环境下基于 Nginx uwsgi 搭建 Django 站点
以下用一个网站 ocean_monitor 举例
MySQL 安装与配置
安装
MariaDB is shipped in the CentOS repo as of CentOS 7 instead of mysql.
if you still want to install mysql you need to add mysql rpm dependency into your yum repo.
sudo yum -y install mariadb-server mariadb-devel mariadb
sudo systemctl start mariadb.service
sudo systemctl enable mariadb.service
配置
登录
mysql -u root -p
创建用户
CREATE USER ocean_monitor IDENTIFIED BY 'ocean_monitor_pwd';
上面建立的用户可以在任何地方登陆。如果要限制在固定地址登陆,比如 localhost 登陆:
CREATE USER ocean_monitor@localhost IDENTIFIED BY 'ocean_monitor_pwd';
创建数据库
# 使用utf8编码,否则中文会有问题
CREATE DATABASE ocean_monitor character set utf8;
授权 ocean_monitor 用户拥有 ocean_monitor 数据库的所有权限
grant all on ocean_monitor.* to ocean_monitor identified by 'ocean_monitor_pwd';
如果是限制在 localhost 登录的,则使用
grant all on ocean_monitor.* to ocean_monitor@localhost identified by 'ocean_monitor_pwd';
使用 PIP 安装依赖
为mysql-python安装依赖,这样才能编译安装
如果是Ubuntu用户
sudo apt-get install build-essential python-dev libmysqlclient-dev python-mysqldb
如果是Centos用户
yum install gcc python-devel
安装所有的依赖
# requirements.txt 是 django 项目目录下,填写的依赖包信息
pip install -r requirements.txt
测试 uWSGU
创建一个测试文件
# test.py
def application(env, start_response):
start_response('200 OK', [('Content-Type','text/html')])
return ["Hello World"] # python2
#return [b"Hello World"] # python3
然后,Run uWSGI:
uwsgi --http :8000 --wsgi-file test.py
如果出现错误,!!! no internal routing support, rebuild with pcre support !!!
sudo apt-get install libpcre3 libpcre3-dev
sudo pip uninstall uwsgi
sudo apt-get remove uwsgi
sudo pip install uwsgi
打开下面url,浏览器上应该显示hello world
http://example.com:8000
如果显示正确,说明下面3个环节是通畅的:
the web client <-> uWSGI <-> Python
测试Django
我看别人用的是 --module mysite.wsgi,但是建 Django项目时,并没有生成这个文件,这里十分奇怪。因此,使用 Django 生成的 wsgi.py
uwsgi --http :8000 --wsgi-file wsgi.py
如果显示正确,说明下面3个环节是通畅的:
the web client <-> uWSGI <-> Django
Nginx 安装与配置
Nginx 安装
http://cantgis.blog.51cto.com/5788192/1540004
配置站点
vim /etc/nginx/conf.d/ocean_monitor.conf
添加配置信息
# ocean_monitor.conf
# the upstream component nginx needs to connect to
upstream django_ocean_monitor {
# server unix:///path/to/your/mysite/mysite.sock; # for a file socket
# for a web port socket (we'll use this first)
server 127.0.0.1:8108;
}
# configuration of the server
server {
# the port your site will be served on
listen 8008;
# the domain name it will serve for
# substitute your machine's IP address or FQDN
# Django 的 settings.py 文件中的 ALLOWED_HOSTS 要加上这里设置的 server_name
server_name localhost;
charset utf-8;
gzip on;
gzip_min_length 1000;
gzip_buffers 4 16k;
gzip_http_version 1.1;
gzip_comp_level 3;
gzip_vary on;
# 禁用对 IE 6 使用 gzip 压缩
gzip_disable "MSIE [1-6].";
gzip_types text/plain text/css text/xml text/javascript application/x-javascript application/xml application/xml+rss application/json;
## Individual nginx logs
access_log /var/log/nginx/ocean_monitor_access.log;
error_log /var/log/nginx/ocean_monitor_error.log;
# max upload size
client_max_body_size 8M; # adjust to taste
# Django media
location /media {
# your Django project's media files - amend as required
alias /home/python/ocean_monitor/media;
}
location /static {
# your Django project's static files - amend as required
alias /home/python/ocean_monitor/static;
}
# Finally, send all non-media requests to the Django server.
location / {
uwsgi_pass django_ocean_monitor;
# the uwsgi_params file you installed
# 增加 nginx 配置, uwsgi_params 文件在 /etc/nginx/ 目录下
include /etc/nginx/uwsgi_params;
}
}
测试 nginx 的配置文件的语法是否正确
sudo nginx -t
重启 nginx
service nginx restart
Django 配置
settings.py 配置
设置 ALLOWED_HOSTS
Django gives Bad Request (400) when DEBUG = False
The ALLOWED_HOSTS list should contain fully qualified host names, not urls. Leave of the port and the protocol. If you are using 127.0.0.1, I'd add localhost to the list too
ALLOWED_HOSTS = [
# 加上本机的IP地址
'192.168.137.146',
'127.0.0.1',
'localhost'
]
You could also use * to match any host:
ALLOWED_HOSTS = ['*']
设置 DATABASES
DATABASES = {
'default': {
'ENGINE': 'django.db.backends.mysql',
'NAME': 'ocean_monitor',
'USER': 'ocean_monitor',
'PASSWORD': 'ocean_monitor_pwd',
'HOST': 'localhost',
'PORT': '3306',
}
}
初始化数据库
python manage.py syncdb
启动一下,测试 Django 有没有问题
python manage.py runserver 0.0.0.0:8080
uWSGI 配置文件
在 Django 项目的根目录添加 uwsgi.ini 文件
# uwsgi.ini file
[uwsgi]
# Django-related settings
# the base directory (full path)
chdir = /home/python/ocean_monitor
# Django's wsgi file
wsgi-file = /home/python/ocean_monitor/ocean_monitor/wsgi.py
# module = index.wsgi:application
# the virtualenv (full path)
# home = /path/to/virtualenv
daemonize = /home/python/ocean_monitor/ocean_monitor.log
# process-related settings
# master
master = true
pidfile = /tmp/ocean_monitor_master.pid
# maximum number of worker processes
processes = 3
# the socket (use the full path to be safe
# socket = /home/python/ocean_monitor/ocean_monitor.sock
socket = 127.0.0.1:8108
# ... with appropriate permissions - may be needed
chmod-socket = 664
# clear environment on exit
vacuum = true
uwsgi 启动与结束
# 启动
uwsgi --ini uwsgi.ini
# 重启
uwsgi --reload /tmp/ocean_monitor_master.pid
# 结束
uwsgi --stop /tmp/ocean_monitor_master.pid
如果出现错误 signal_pidfile()/kill(): No such process [core/uwsgi.c line 1627],是由于 ocean_monitor_master.pid 的进程ID不正确,修改/tmp/ocean_monitor_master.pid为正确的pid就可以。
使用如下命令,查询指定端口的进程id
sudo netstat -ap | grep 8108
修改 /tmp/ocean_monitor_master.pid 的进程id值
vim /tmp/ocean_monitor_master.pid
使用 supervisor 启动 celery-worker 及 celery-beat
安装 supervisor
安装
pip install supervisor
安装方法请参考
http://www.iitshare.com/supervisord-manage-process.html
生成配置文件
安装好supervisor之后,默认是没有生成配置文件的。可以通过以下命令生成配置文件
echo_supervisord_conf > /etc/supervisord.conf
启动
supervisord -c /etc/supervisord.conf # 到指定路径下去找配置文件
如果出现
another program is already listening on a port that one of our HTTP servers is configured to use
sudo unlink /etc/supervisor.sock
然后再次运行
supervisord -c /etc/supervisord.conf
supervisor 管理
supervisor 安装完成后有两个可用的命令行 supervisor 和 supervisorctl,命令使用解释如下:
supervisord,初始启动Supervisord,启动、管理配置中设置的进程。
supervisorctl stop programxxx,停止某一个进程(programxxx),programxxx为[program:chatdemon]里配置的值,这个示例就是chatdemon。
supervisorctl start programxxx,启动某个进程
supervisorctl restart programxxx,重启某个进程
supervisorctl stop groupworker: ,重启所有属于名为groupworker这个分组的进程(start,restart同理)
supervisorctl stop all,停止全部进程,注:start、restart、stop都不会载入最新的配置文件。
supervisorctl reload,载入最新的配置文件,停止原有进程并按新的配置启动、管理所有进程。
supervisorctl update,根据最新的配置文件,启动新配置或有改动的进程,配置没有改动的进程不会受影响而重启。
注意:显示用stop停止掉的进程,用reload或者update都不会自动重启。
设置配置文件
vim /etc/supervisord.conf
在文件最后添加如下信息
[program:ocean_monitor_celery_worker]
process_name = ocean_monitor_celery_worker
command=python /home/python/ocean_monitor/manage.py celery worker --loglevel=info
directory=/home/python/ocean_monitor
autorestart=true
redirect_stderr=true
stdout_logfile = /var/log/supervisord/ocean_monitor_celery_worker.log
loglevel=info
[program:ocean_monitor_celery_beat]
process_name = ocean_monitor_celery_beat
command=python /home/python/ocean_monitor/manage.py celery beat --loglevel=info
directory=/home/python/ocean_monitor
autorestart=true
redirect_stderr=true
stdout_logfile = /var/log/supervisord/ocean_monitor_celery_beat.log
loglevel=info
用 supervisor 运行前,可以先在命令行下启动测试一下
python manage.py celery beat --loglevel=info
python manage.py celery worker --loglevel=info
启动
# 根据最新的配置文件,启动新配置或有改动的进程
# 配置没有改动的进程不会受影响而重启
supervisorctl update
防火墙设置及 SELinux 问题
设置防火墙开放端口(iptables)
/sbin/iptables -I INPUT -p tcp --dport 8008 -j ACCEPT
nginx permission denied 问题
因为忘记了 selinux 的问题,导致 trouble shooting 耽误了不少时间。
由于SELinux的问题,会导致nginx nginx permission denied 的问题,需要将 selinux 关闭。
查看SELinux状态:
# 如果SELinux status参数为enabled即为开启状态
/usr/sbin/sestatus -v
也可以用 getenforce 这个命令检查
关闭SELinux
# 只是临时关闭(不用重启机器),重启后问题仍然出现
# 设置SELinux 成为permissive模式
setenforce 0
# setenforce 1 # 设置SELinux 成为enforcing模式
永久关闭,需要通过修改配置文件,但是需要重启机器:
vim /etc/selinux/config
将SELINUX=enforcing 改为 SELINUX=disabled,重启机器即可
shutdown -r now