Servere server problem (desperate!!)

[Factorhost]

Hi everyone.

Well, I run a forum with 20,000+ userbase
I'd say at peak 200-300 users online at once..

My server is a quad core, 4gb RAM machine, running cPanel of course

Not many accounts on this box by the way.

PROBLEM:

Since we colo'd this box, every 7-or-so days, the server would just not respond. No errors, nothing. Just won't respond. Dead. (requests timed out)

So we'd get a reboot, and this would happen again and again.

So then, I realised mysql wasnt configured for all our lovely cores and ram, so i did that. Hurray! i thought, 12-15 days uptime, load never went above 1.0.

Then, today, it happened again..


Mite i note, i check logs once server is rebooted. Nothing, no error, no warn, no kernel panic.

Please someone help, this is seriously ruining me, if its not software, its gonna be a new server

[Infopro]

Forum up to date? Many Hacks/Mods installed to it? Good places to start. Check your error log on the account for problems that can be fixed.

Install this is you don't have it already and set it up to kill off runaway processes. http://www.configserver.com/cp/csf.html

What version of Apache? Mod_Security installed?

[Factorhost]

Thanks for quick reply.
Yeah the forum is latest (VB)
The mods are fine, I'm not even sure if the issue is MySQL anymore?

We have CSF, to be honest, I'm starting to question if it is that (crashing, blocking or something?)

[sirotex]

Hello,

Your problem could be heaps of things, like someone DDoSing your server or even dodgy hardware.

But if the Server is still up and just not responding more than likely someone running a DDoS

[rpmws]

Hi everyone.

Well, I run a forum with 20,000+ userbase
I'd say at peak 200-300 users online at once..

My server is a quad core, 4gb RAM machine, running cPanel of course

Not many accounts on this box by the way.

PROBLEM:

Since we colo'd this box, every 7-or-so days, the server would just not respond. No errors, nothing. Just won't respond. Dead. (requests timed out)

So we'd get a reboot, and this would happen again and again.

So then, I realised mysql wasnt configured for all our lovely cores and ram, so i did that. Hurray! i thought, 12-15 days uptime, load never went above 1.0.

Then, today, it happened again..


Mite i note, i check logs once server is rebooted. Nothing, no error, no warn, no kernel panic.

Please someone help, this is seriously ruining me, if its not software, its gonna be a new server

can you get a console terminal output started and wiat for this to happen? if so ..see if the errors are journal commit errors or something related let us know.

[Factorhost]

can you get a console terminal output started and wiat for this to happen? if so ..see if the errors are journal commit errors or something related let us know.

I can't even ping.
It's just totally unresponsive....


I've just got it back up, and literally the last log entry before it happened is minute and nothing to do with a crash. After it is the boot message from today.


:S This really is bugging me...
Any more ideas?

[jahja]

Well, I run a forum with 20,000+ userbase
I'd say at peak 200-300 users online at once..

My server is a quad core, 4gb RAM machine, running cPanel of course

Have you calculated your RAM usage? You can just ps auxf, and calculate the sum of RSS. If you say 200-300 users online at once, it will easily 6 GB to 9 GB. If your swap file is 4 GB, then certainly, it will hang once 300 users are online.

The solution is either to add the RAM, or add swap file, or change to lighttp, or optimize apache and mysql such as limit the apache instances that can be called, or optimize SQL query so that it will work faster to serve users (reducing users online at the same time).

I had one VPS customer that faced the same problem and he chose to add RAM and solved his problem.

Jahja.
OmniVPS.com
Bataviahost.com