Hi to all,
has anybody tried to run Webspeed with Ngnix-Webserver ?
We look for an Server with high throughput, but Ngnix does not support cgi out of the box, so can anybody help how to implement a secure and fast cgi-scripting environment ?
We don´t use PAS yet so we need this kind of Webspeed-Services.
Thanks in advance.
Nginx no, Lighttpd yes, and haproxy.
Have a test system set up at work for a Webspeed that runs on Debian 8 instead of RHEL.
User goes to the website via the HAproxy port (running on port 80), depending on the request, if it is just for a standard web request it gets pushed over to LightTPD+CGI+WebSpeed port, if it is for websocket it gets pushed over to the redis-server port.
IMHO I find LightTPD to be much faster than Nginx, but that's just personal preference really.
This is interesting. I am interested in how you set-up LightTPD working with WebSpeed. In theroy it should work on Windows Server...right?
I scrapped the HAproxy side of it since I decided to do websockets a different way, but the lighttpd side of things has been working quite nicely (and so much faster than Apache).
For lighttpd, I am just running a stock standard setup (with the exception of the cgi-bin side of things).
lighttpd.conf:
server.modules = (
"mod_access",
"mod_accesslog",
"mod_alias",
"mod_compress",
"mod_redirect",
"mod_setenv",
)
server.document-root = "/var/www"
server.upload-dirs = ( "/var/cache/lighttpd/uploads" )
server.errorlog = "/var/log/lighttpd/error.log"
accesslog.filename = "/var/log/lighttpd/access.log"
server.pid-file = "/var/run/lighttpd.pid"
server.username = "www-data"
server.groupname = "www-data"
server.port = 80
# Keep-Alive stuff
# Was at 32 and 3
server.max-keep-alive-requests = 0
server.max-keep-alive-idle = 3
index-file.names = ( "index.php", "index.html", "index.lighttpd.html" )
url.access-deny = ( "~", ".inc" )
static-file.exclude-extensions = ( ".php", ".pl", ".fcgi" )
compress.allowed-encodings = ("gzip", "deflate")
compress.filetype = ( "application/javascript", "text/css", "text/html", "text/plain" )
10-cgi.conf (in /etc/lighttpd/conf-enabled)
server.modules += ( "mod_cgi" )
$HTTP["url"] =~ "^/cgi-bin/" {
cgi.assign = ( "" => "" )
}
alias.url += ( "/cgi-bin" => "/usr/lib/cgi-bin/" )
Then in the /usr/lib/cgi-bin directory I just have my cgi script for my broker (eg. cgiip) which points to the host the DB is running on (that also has the actual broker and agents on it).
I had to reduce the max-keep-alive-requests from the default as having it at 32 was causing issues.
Response times (according to 'wtbman -i broker' have basically gone from 200ms or so down to about 10ms just by switching to lighttpd).