Apache leading to system lock/freeze

The conclusion in the title may be wrong given that the problem is close to a year old and just now was I able to get some info on the issue. This is on a centos 5, running on an old PIII with 384 MB Ram. Note that before about a year ago, when I don't recall what exactly changed, this same server was working just fine. It may have been an older CentOS on it though.

uname says "Linux localhost.localdomain 2.6.18-410.el5.centos.plus #1 SMP Wed May 11 21:46:57 EDT 2016 i686 i686 i386 GNU/Linux"

The problem presents itself with random freezes, every few days, sometimes twice a day. The freeze is not a real freeze in that if I try to connect with ssh to the server, after a good few minutes, I do get a user prompt and after another very long time, I get disconnected with timeout, so I don't get to enter the password. But eventually, if I let it "frozen" for a day or two, it completely lock up.

The only solution is a reboot, from the button.

Yesterday I finally got a hint from the physical server in that the HDD led was fully lit at the time of the freeze (didn't notice that before as I wasn't really looking), so I fired up IOTOP to monitor the activity and left it running in a putty window.

Just now it froze up at some point. Here is the screenshot.

I see there are a lot of httpd processes, a couple mysqld and the very top one is the jualing.

I have no idea how to interpret those percentages in the context of my server freezing up.

I have a strong feeling this is a configuration problem and I should be able to control this by tweaking some configurations but am failing to guess what that is.

I do know that restarting apache every few hours prolongs the time the server is ok. And if my traffic goes up (from a very short few visits a day), then the server gets to freeze up in a matter of hours.

Any pointers in what else to check? Or what configuration to tweak?

Thank you.

Recent Questions...