12 messages in ru.sysoev.nginxRe: Nginx isn't as fast as i expect
FromSent OnAttachments
somebody nobodyAug 16, 2007 3:15 am 
Igor ClarkAug 16, 2007 3:37 am 
Igor SysoevAug 16, 2007 4:03 am 
Brice FigureauAug 16, 2007 4:20 am 
somebody nobodyAug 16, 2007 4:21 am 
somebody nobodyAug 16, 2007 4:31 am 
seun...@public.gmane.orgAug 16, 2007 4:46 am 
Brice FigureauAug 16, 2007 4:52 am 
seun...@public.gmane.orgAug 16, 2007 5:00 am 
JoeyAug 16, 2007 7:45 am 
Aleksandar LazicAug 17, 2007 1:18 am 
seun...@public.gmane.orgAug 18, 2007 11:21 pm 
Actions with this message:
Paste this link in email or IM:
Paste this link in email or IM:
Atom feed for this thread
Paste this URL into your reader:
Subject:Re: Nginx isn't as fast as i expectActions...
From:Brice Figureau (bric@public.gmane.org)
Date:Aug 16, 2007 4:20:16 am
List:ru.sysoev.nginx

On Thu, 2007-08-16 at 12:16 +0200, somebody nobody wrote:

user nobody nogroup; worker_processes 10;

uh 10 ? That sounds like a lot to me. I think you should not use more than 1 worker_process per processor or core.

sendfile off;

I don't know if solaris has sendfile but usually sendfile is faster (or more resource friendly) than using read/write combinations.

location ~ \.php$ { include /nginx/conf/fcgi.conf; fastcgi_pass unix:/tmp/fastcgi.socket; fastcgi_index index.php; }

Are you sure Apache is using the same FastCGI settings as nginx ones? How are you spawning your fastcgi process? If you are using lighttpd's spawn-cgi make sure you have the same number of PHP processes as what you are using with Apache. Did you try a tcp socket instead of a unix one ?

Can you give us the ab parameters you used, especially the concurrency ?

-- Brice Figureau <brice+ngin@public.gmane.org>