<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"></head><body ><div>Having one block per server is the best for performance because you only test a condition once; the way you're doing it would require NGINX to evaluate hostnames twice or maybe more.</div><div><br></div><div>In my opinion config maintenance is important as well so either stick to what you have (unless those sites have thousands of requests per second) or generate the configs using Puppet, Chef or the like...</div><div><br></div><div><br></div><div><br></div><div><br></div><div><div style="font-size:75%;color:#575757">Sent from Samsung Mobile</div></div> <br><br><br>-------- Original message --------<br>From: Darren Pilgrim <list_nginx@bluerosetech.com> <br>Date: <br>To: nginx@nginx.org <br>Subject: Converting Apache configs to nginx, why is a NameVirtualHost workalike is a bad thing? <br> <br><br>I'm switching some servers from Apache 2.2.x to nginx 1.2.6. In Apache, <br>I use NameVirtualHost. I also use Rewrite directives to manage www. <br>prefixing via a 301 redirect. All the user has to is create <br>/wwwroot/$hostname, upload the site files, and point DNS at my server. <br>On my end, all I have to do is create the Apache config once as part of <br>the new-user setup.<br><br>When I converted that to nginx, this is what I came up with:<br><br>root /www/user/wwwroot/$http_host;<br><br>For the user who wants to drop the www.:<br><br>if ($http_host ~ ^www\.(.+)$) {<br> return 301 http://$1$request_uri;<br>}<br><br>For the user who wants to always have the www.:<br><br>if ($http_host !~ ^www\.) {<br> return 301 http://www.$http_host$request_uri;<br>}<br><br>Which, in testing, works a treat. Various articles say this style of <br>configuration is bad. Instead I should:<br><br>1. have per-domain server blocks;<br>2. have a server block for www.example.com that redirects to <br>example.com. or vice versa.<br><br>In other words, for the user who has 37 domains, I'll need 74 server <br>blocks in their nginx config. That's a significant regression in terms <br>of workload and simplicity.<br><br>If it's bad, ok, I won't do that; however, I can't seem to find an <br>explaination *why* it's bad. Would someone please clarify that point?<br><br>_______________________________________________<br>nginx mailing list<br>nginx@nginx.org<br>http://mailman.nginx.org/mailman/listinfo/nginx<br></body>