up vote -7 down vote favorite
share [g+] share [fb]

I'm a web designer by trade who develops using php and mysql. I have this wordpress-powered website running alone on a dedicated server (centOS 5.7, php5.3.8 + APC , Apache 2.2, mysql 5.1 - hardware: intel celeron 2.8Ghz, 2go RAM). I manage the server via cPanel WHM. I've installed munin to visually view the strain on each

In terms of traffic: its always around 10 to (max) 50 concurrent users, average around 20.

Every day, the server crashes once. I fail to find the problem. I would need some guidance in order to understand why exactly the server resources go 100% which leaves me no other choice than to hard-reboot.

SYMPTOM DESCRIPTION

"Crash" = server becomes unresponsive, munin tells me: cpu = 100%, ram=100%, swap=100%.

The crash happens every day but not at the same time. The server is brand new, we only have it since one month and it's taking the real traffic only since December 5th (DNS redirection happened on November 28th, then some advertising occured on monday). Since Dec 5th, it crashed once or twice a day. I found out page caching by a Wordpress plugin (W3 Total Cache) was not working correctly and it seemed to solve the problem. Then it crashed yesterday (6 Dec) after i desactivated it.

Today it crashed twice. But i noticed in the error_log there were still requests for W3 total Cache object-cache.php so i completely deleted the plugin 's files.

The Munin graphs show that it happens very fast. Everything is ok (cpu, ram, swap) for hours, then all of a sudden memory, cpu and swap rise at the same time. Unfortunately munin does not give access to the exact list of processes that would provoke the spike.

How do you audit a failing LAMP system? What steps should i take?

Thank you for your advises. I realise i'm beyond my field of expertise, but i really do enjoy the few things i've learned to do so far via ssh and i'd like to know more from real experts, what to do in such situations. Knowledge never hurts and will probably help me produce better php code.

link|improve this question
57% accept rate
Did you check your system logs /var/log/messages and /var/log/syslog? – Khaled Dec 7 '11 at 12:45
9  
If you're inexperienced why aren't you on managed hosting? – sam Dec 7 '11 at 12:46
4  
sam wasn't being sarcastic nor was he showing off (at least doesn't appear to be). This is a completely legitimate question. You have no business managing a client's internet-facing dedicated server if you are an inexperienced systems administrator. This is what managed hosting services are for. – Doug Dec 7 '11 at 13:33
1  
Wow, talking back to users asking legit questions isn't exactly a way to solicit free assistance from people. I'd downvote just for the unwillingness to work with people who are offering assistance for free... – Bart Silverstrim Dec 7 '11 at 13:39
1  
I don't really care one way or the other. I was more trying to point out to you that you should holster the ego when asking people for some advice or else they'll brush you off as a jerk and not feel any need to invest time in helping you out. – Bart Silverstrim Dec 7 '11 at 14:02
show 6 more comments
feedback

1 Answer

up vote 1 down vote

You didn't say who hardware vendor for the server so I am going to assume it is a "beige box".

Faults of these kind usually indicate some kind of failing hardware.

First I would check the memory using memtest86 and make sure that any memory modules are okay. Also if you can, crack open the server and make sure that dust isn't causing any problems.

Then I would then check the harddisk using fsck and to check that it is not failing.

Other than I would suggest running updates to make sure that you are running the latest versions of the Linux Kernel, php, MySql, etc.

link|improve this answer
If it's a hardware problem on a traditional rented dedicated server (most likely), the vendor doesn't matter, he should just be going to the data center he's renting from for all hardware related requests. I'm doubting he has access to physically crack open the server. – Doug Dec 7 '11 at 13:37
@Doug is correct, it is a dedicated server rented on kimsufi.com. – pixeline Dec 7 '11 at 13:39
3  
So the answer to your problem, if you don't have the experience or ability to troubleshoot this problem yourself and you believe it is a hardware issue, is to go ahead and open a ticket with the data center you are hosting with and see if they'll do some hardware diagnostics for you. Granted, it is likely the server will be taken off line while they are doing so. – Doug Dec 7 '11 at 13:42
All right, will do that. Just told the customer it's just not my line of competence. – pixeline Dec 7 '11 at 14:01
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.