Need help optimizing web server.

2025-04-07

I have a web server running

Athlon XP1800+
1G RAM
2x 60G IDE harddisk running RAID 1
RedHat Linux 7.3
Apache 1.3.27
MySQL 3.23.56
PHP 4.1.2 + Turck MMCache

It hosted a few web sites, but the busiest one is a forum running Invision Power Board (IPB). It usually has 100-150 users logging in at the same time. There are nearly 1 million posts. It works very well when it is running IPB 1.1. However, after a recent upgrade to IPB 1.2, the server become very slow. The main problem is there are excessive hard disk activity. Without the hard disk activity, the php can be loaded in 0.05 - 0.20 seconds. But the harddisk activity is so frequent that some times pages takes more than 10 seconds to load. Even worse it ends up in too many SQL conection frequently. The hard disk activity last for 30-60+ seconds from time to time.

A free command show that there should be enough memory.

[root@jupiter root]# free
total used free shared buffers cached
Mem: 1031388 991364 40024 0 20672 617236
-/+ buffers/cache: 353456 677932
Swap: 1044216 18448 1025768

I have try to manipulate the my.cnf of mysql. I try to use my-large.cnf, to my astonishment, the system start using up the swap instead of the cache. I change back to my-medium.cnf it stops using the swap, but there are still a lot of hard disk activity. I have checked the cron job and everything else and try to stop unnecessary programs. But the problems seems to lie in either the httpd or mysqld.

Could somebody enlighten me where is the problem? What is moving my hard disk? How can I optimise my web server, mySQL or php?