|
Lighttpd performance problem + RAID performance problem in a high load site
Hi!
I´m running a Music site with flash players for play MP3 files. We have more or less 250.000 visits per day with 5-10 req/s and 4 servers (Mysql, Web, Store, Backup).
Web server specs:
Xeon Dual Core 3065
4 GB RAM DDR2
2x146 GB SAS 15.000 RPM
Linux Centos 5.2 64 Bits
Lighttpd 1.4.20 (RPM)
PHP 5.1.6
Access to MP3 files via NFS (Gigabit Ethernet link)
Load average: 4.5 - 6 (Some times 120 :x)
Store server:
Xeon E5410
8 GB RAM DDR2
2x146 GB SAS 15.000 RPM (RAID 1) (SYSTEM)
6X1000 GB SATA 2 (RAID 6) (MP3 store)
Linux Centos 5.2 64 Bits
NFS V4 daemon
CPU Usage: 1-5 %
Load average: 6-7
RAID Controller: ADAPTEC SAS RAID 3805 8 Ports PCI-EXPRESS 128 Mbit DDR2 (connected SAS and SATA disks)
Lighttpd Configuration:
server.modules = (
"mod_rewrite",
"mod_redirect",
"mod_access",
"mod_fastcgi",
"mod_accesslog" )
server.document-root = "/var/www/website/"
server.max-connections = 7000
server.max-fds = 21000
server.max-worker = 30
#server.max-request-size = 30000
server.max-write-idle = 7200
server.max-read-idle = 10
server.max-keep-alive-requests = 0
server.max-keep-alive-idle = 0
#server.use-noatime = "enable"
server.event-handler = "linux-sysepoll"
server.network-backend = "linux-sendfile"
fastcgi.server = ( ".php" =>
( "myip" =>
(
"socket" => "/var/run/lighttpd/php-fastcgi.socket",
"bin-path" => "/usr/bin/php-cgi",
"max-procs" => 10,
"bin-environment" => (
"PHP_FCGI_CHILDREN" => "16",
"PHP_FCGI_MAX_REQUESTS" => "5000"
)
))
)
I got a lot of this errors and php-cgi zombie process:
2008-11-12 19:54:48: (mod_fastcgi.c.2956) backend died; we'll disable it for 5 seconds and send the request to another backend instead: reconnects: 0 load: 9
2008-11-12 19:54:48: (mod_fastcgi.c.2709) child died somehow, waitpid failed: 10
Connections at this moment:
# netstat -an |grep :80 |wc -l
2409
# netstat -an |grep :80 |grep TIME_WAIT |wc -l
1102
Could be the raid causing troubles mixing disks SATA/SAS and with a RAID6 implementation (writing slow) in my store server?
Is linux-sendfile (rpm default backend) causing troubles with NFS?
And the most important question: is my configuration correct?
I appreciate any help.
Thanks!
|