uWSGI request timeout in Python

问题:

Trying to set the timeout for requests in uWSGI, I'm not sure of the correct setting. There seem to be multiple timeout options (socket, interface, etc.) and it's not readily evident which setting to configure or where to set it.

The behavior I'm looking for is to extend the time a request to the resource layer of a REST application can take.

回答1:

You're propably looking for the harakiri parameter - if request takes longer than specified harakiri time (in seconds), the request will be dropped and the corresponding worker recycled.

For standalone uwsgi (ini config):

[uwsgi]
http = 0.0.0.0:80
harakiri = 30
...

If you have nginx proxy before uwsgi you have to increase timeout as well:

  location / {
    proxy_pass http://my_uwsgi_upstream;
    proxy_read_timeout 30s;
    proxy_send_timeout 30s;
  }

If you want (for some strange reason) higher timeout than 60s, you might consider communication over uwsgi protocol. Configuration is quite similar nginx site:

location / {
    uwsgi_read_timeout 120s;
    uwsgi_send_timeout 120s;
    uwsgi_pass  my_upstream;
    include     uwsgi_params;
}

uwsgi:

[uwsgi]
socket = 0.0.0.0:80
protocol = uwsgi
harakiri = 120
...

回答2:

Setting http-timeout worked for me. I have http = :8080, so I assume if you use file system socket, you have to use socket-timeout.