Deploy web using gunicorn Py application

I often see people ask python Of web How to deploy the application .

A common solution is to use apache Of mod_wsgi, Or a little high-end uWsgi. But personally, I think these are too complicated , It is inconvenient to use and maintain , I usually recommend gunicorn, coordination Apache or Nginx, It can realize simple and efficient application deployment and maintenance .

For this reason, I specially wrote such a document to introduce , And this article submits pull request To web.py Project cookbook, It's just Aaron Swartz After death ,web.py I don't know who is maintaining , It was submitted for a month and no one handled it , Look at the pull request list , also 2 The request years ago is hanging ……

So I had to fork One. , Add this page , The maintenance I will do in the future , Will be updated on this branch .

adopt Gunicorn stay Apache and Nginx Deploy

The following code is based on Gunicorn 0.14.6 stay Debian 6.0.6 and FreeBSD 9.0 Test under the system .

The latest version of Gunicorn There is no difference in the method of use . Other Linux Distribution version 、 Various BSD、Mac OS X The system should be OK . But it is not recommended. Windows System , Because there are many unnecessary difficulties in visual inspection .

Note:

• You can rename code.py For any name you like , This example is based on code.py For example .
• /path-to/webpy-app To include your code.py The path of the code .
• /path-to/webpy-app/code.py It should be yours **python file** Full path to .

You can run... On the command line gunicorn --version View the current gunicorn Version of .

install Gunicorn

See Gunicorn Official website ：

http://gunicorn.org/#quickstart

The suggestion on the official website is to use virtualenv Way to install , It's a good way . I won't introduce virtualenv（ Personal recommendation virtualenvwrapper）, The following is the installed virtualenv Environment as an example , Of course, you can also skip virtualenv, Direct global installation （ Need to add sudo）.

pip install gunicorn

Copy

use Gunicorn Deploy web.py application

Gunicorn It's for deployment wsgi Applied , Any support wsgi All applications are OK , It's not just web.py.

The whole deployment process is divided into two parts ：

• use Gunicorn function web.py/wsgi application
• To configure web server Front end reverse proxy

use Gunicorn function web.py application

I've said that before ,Gunicorn It's used to deploy wsgi Applied , So first of all, we need to modify code.py, Make it one wsgi application .

#  ...
app = web.application(urls, globals())
#   Add the following sentence here , that will do 
application = app.wsgifunc()

Copy

The simplest way to run it is ：

gunicorn code:application

Copy

among code Is refers to code.py,application That's it. wsgifunc Name .

If it works like this , gunicorn It is used as a listener by default 127.0.0.1:8000 Of web server, It can be accessed locally ： http://127.0.0.1:8000 visit .

If you want to access through the network , You need to bind different addresses （ You can also set the listening port at the same time ）：

gunicorn -b 192.168.0.123:8080 code:application

In multicore The server On , In order to support more concurrent access and make full use of resources , You can use more gunicorn process ：

gunicorn -w 8 code:application

So you can start 8 Two processes at the same time HTTP request , Improve the efficiency and performance of the system .

in addition , gunicorn The default network model is synchronous blocking (-k sync), It may not perform well enough for large concurrent accesses , It also supports other better patterns , such as ：gevent or meinheld.

#  gevent
gunicorn -k gevent code:application
#  meinheld
gunicorn -k egg:meinheld#gunicorn_worker code:application

Copy

Of course , To use these two things, you need to install them separately , Please refer to the respective documents for details .

The above settings can also be set through -c Parameters are passed into a configuration file to implement .

To configure Apapache Reverse proxy of

The simple reverse proxy configuration is as follows （ In the VirtualHost Li as an example ）：

    ProxyPass / http://127.0.0.1:8000/
    ProxyPassReverse / http://127.0.0.1:8000/
    ProxyPreserveHost On
    ProxyErrorOverride Off

Copy

Proxy all access requests to the root path to http://127.0.0.1:8000 Of gunicorn Service .

To configure Nginx Reverse proxy of

The simple reverse proxy configuration is as follows （ The same is true of virtual host For example ）：

    location / {
        try_files $uri @test;
    }

    location @test {
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $http_host;
        proxy_pass http://127.0.0.1:8000;
    } 

Copy

Proxy all access requests to the root path to http://127.0.0.1:8000 Of gunicorn Service .

In practical application, you may need to set more proxy_set_header Variable , It depends on the application requirements .