Nginx & long poll: Best practices to reduce memory and bandwidth footprint

[Thomas Love]

On 4 November 2011 05:15, Quintin Par <quintinpar at gmail.com> wrote:

> Posting the contents inline. Can someone review this?
>
> -----------------------------------------------------------------------------------------------------
> I use nginx in this mode for [BOSH][1] and chat clients along with gzip.
>
> location ~* /http-bind/ {
>      proxy_buffering off;
>      keepalive_timeout 55;
>      access_log off;
>      tcp_nodelay on;
>      proxy_pass http://x.x.x.x:1111;
>  }
>
> Is this the best approach to **managing long polling** in nginx.
>
> I also use just one worker process for altogether for web & chat (single
> CPU). Is that fine?
>
>
It looks fine. Use that until you have a performance problem, and then
gather evidence identifying your bottleneck before changing anything.

You might as well start with n worker processes though, where n is the
number of cores on your CPU. Give yourself a few thousand
worker_connections, because you'll have a relatively large number of
relatively idle sockets.

Long-polling is very simple and nginx is good at it. I would advise not to
try to fix anything until you find a real problem.

Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx/attachments/20111104/2d8a7165/attachment.html>