My server became overloaded today

9:17am up 1 day, 7:07, 1 user, load average: 4.34, 5.80, 5.68
455 processes: 448 sleeping, 7 running, 0 zombie, 0 stopped
CPU states: 70.8% user, 5.5% system, 0.0% nice, 23.6% idle
Mem: 1028456K av, 1003716K used, 24740K free, 1352K shrd, 202500K buff
Swap: 1024088K av, 604K used, 1023484K free 509552K cached

I notice some httpd processes take more that 10% CPU.
Is this normal?
How to reduce the load without decreasing the performance?

We would need more information. If you have a server-status page set up you can tell the number of requests that were cokming into the server, that would help determine if the CPU usage is simply because there are a lot of requests -- or because of something else.

In addition some modules, like mod_gzip or mod_perl, can add to the HTTP CPU usage, but this increase in CPU usage offsets CPU a potentially larger CPU usage if those modules were not installed.

uuallan,

Thank you for you answer.

Unfortunately I don't have such a page and don't know how to set up one :(

In fact there are some perl scripts that are running on the server but I try to start them one by one not to overload it.
May be there needs to be done some perl optimization.
I set MaxRequestsPerChild to 100 if this helps at all.

Would you recommend me an online guide how to do it or describe it here if it is not that hard.

I am a total newbie at linux and perl and am having difficulties understanding it. So please try to put it simple :)

Thanks again!!!

HELPPPPPPPPPPPPPPPPPP!!!!!!!!!!!!!!!!!!!!!!!!!
It's getting worse


11:55am up 4 days, 9:45, 1 user, load average: 37.46, 27.89, 29.95
507 processes: 453 sleeping, 54 running, 0 zombie, 0 stopped
CPU states: 90.0% user, 10.0% system, 0.0% nice, 0.0% idle
Mem: 1028456K av, 1023348K used, 5108K free, 1352K shrd, 130292K buff
Swap: 1024088K av, 652K used, 1023436K free 530088K cached

Now the SMTP has stopped working and won't start.
Please heeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeelp

Take a look at this, and see if it helps:

http://www.linux-mag.com/2001-03/tech_support_03.html

The important thing is to find out which site on the server is causing these the CPU usage.

I reduced Minspareserver and Startservers to 32
Maxspareservers to 64
and Maxclients to 512
But it didn't help much
SMTP still can't start

2:28pm up 2:11, 1 user, load average: 25.54, 19.70, 24.35
441 processes: 410 sleeping, 31 running, 0 zombie, 0 stopped
CPU states: 87.7% user, 7.3% system, 0.0% nice, 4.9% idle
Mem: 1028456K av, 964340K used, 64116K free, 1352K shrd, 104648K buff
Swap: 1024088K av, 0K used, 1024088K free 600004K cached

Is it normal to have more than 30 processes running simultaneously?
How to reduce the number of processes running?

What error do you get when you try to start SMTP?

Were you able to determine which site/what pages were generating the traffic?

I have at least 50 processes like this
apache 10088 0.2 0.8 15232 8980 ? S 00:07 0:18 /usr/sbin/httpd -f /etc/httpd/conf/httpd_app.conf -DSSL -DSSL.d
and lots of log processes:
root 4155 0.0 0.0 1516 416 ? S Jun03 0:00 /usr/local/sbin/cronolog /home/virtual/site44/fst/var/log/httpd

The problem with ps -aux is that I can't see anything that eats more than 1% CPU and every time on Top there is at least one 7-15% httpd

btw, I managed to start the smtp but cut the timouts to 5 secs and the average loads now are below 10