Optimising Dedicated Server for Shared Hosting?

[Forcerdj]

Hi,

So i've recently changed hosts and upgraded my server so that i can delve into the world of web hosting... all was going good until tonight i noticed some sites unresponsive, i couldn't get into WHM so i logged i via SSH which was slow, the monitoring tool said CRITICAL as server load was through the roof.

I restarted Apache from the command line and everything normalised in seconds.

I don't want to start advertising and getting clients when server could spike at anytime and knock websites offline. What happens if sever hits CRITICAL again and we're not around to correct it.

General server usage at the moment is extremely low, as of writing this load is 0.2 (8 CPUs) and memory 16.17%, Swap 7.69%.

I'm wondering how i can avoid this in the future? Optimise the server so that it can better handle the load, or put something in place that kills the process if it gets to dangerous levels?

Any help appreciated, thanks

 

[vanessa]

Perhaps you'll want to first see what the cause of the load is. Next time it gets like this, SSH in and run 'top' to see if anything sticks out. The Daily Process Log in WHM might also show you what's going on.

 

[Infopro]

When load spikes like that, there's a reason for it. Optimizing services is certainly helpful, but your server can still run into trouble regardless of how well its optimized if there's any accounts on the server running out of date scripts and possibly being attacked, for example, a very common thing these days.

You may find this troubleshooting thread useful:
Troubleshooting high server loads on Linux servers - cPanel Forums

My guess, somethings wrong with one of your sites, somewhere.

Do you have ConfigServer Firewall installed? Properly configured, that will give you lots of help keeping things on your server in line, and will alert you by mail about many things on a regular basis.

 

[Forcerdj]

Thanks for the replies, the offending website is an old installation of vBulletin so it's highly likely it's poorly optimised or a is a big target for attacks.

It's a large site with 3GB database so upgrading has always been brushed aside but i think i will bite the bullet over the weekend.

This morning i had a similar sever load issue, again with the vbulletin account.

Time:          Fri Feb 28 02:46:23 2014 +0000
Account:       vbulletin
Process Count: 50 (Not killed)

Process Information:

User:vbulletin PID:27145 PPID:26099 Run Time:42(secs) Memory:272848(kb) exe:/usr/bin/php cmd:/usr/bin/php /home/vbulletin/public_html/photos/displayimage.php
...x50