Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have a 24 Core Opteron 8435 system with 48GB of RAM. However, the performance is way below that of our 6 Core Intel with 30GB of RAM. I wonder if anyone could this of why? We're running a Drupal website off this box with mysql 5.0 installed on the same box. PHP 5.6.13 and Apache 2.2.9. Every now and then the response is quicker than lightening but most times it struggles. mpstat -P ALL shows little activity but it's almost like the apache children are having problem starting up but once started are fine? Is this a CPU cache issue?

My dmesg is at

share|improve this question

migrated from Jul 4 '10 at 0:05

This question came from our site for professional and enthusiast programmers.

Not programming related. Try – tur1ng Jul 3 '10 at 8:35
I found Linux happily put all my Folding At Home processes on one processor core. top, f, j, enter, to see CPU numbers. (The kernel preserves CPU affinity, and doesn't appear to rebalance.) If it is wasting cores, you can use taskset to fix it. – sarnold Jul 3 '10 at 8:58
Unfortunately seems to have died. taskset is interesting but when Apache spawns children will these also need to be taskset I wonder. I'll play tonight and report back if this question is still here. I posted yesterday but it seems to have been removed without notice! ;-) – Khushil Jul 3 '10 at 9:12
@tur1ng It's ! – AndrejaKo Jul 3 '10 at 9:29
possible duplicate of 24 Core Server Performance – sybreon Jul 4 '10 at 2:28

This machine is not in production as yet and is in test and our hosting facility so no problems with making changes/reboots etc.

I have turned off BIOS Memory Interleaving which has indeed made a difference but as yet still about 100ms behind the 12 core box on ab -c 10 -n 10.

I can see the process' moving better across the cores now too.

share|improve this answer

What you need to do is produce a synthetic load in your lab on a similar machine, and compare its behaviour with another one where the problem does not occur.

If you deployed this into production without trying it in a test environment, you can expect trouble.

The idea of running MySQL on the same machine as a web server for a high performance environment troubles me slightly ... you'll get better performance with a small load but it will generally scale less. However your experience may vary.

The key is to reproduce the environment exactly and run load testing against it, this is not necessarily easy, but it's the only way to know what's going to happen before getting it into production and having it affect your users.

share|improve this answer

Did you make any tweak at all? otherwise I think you can only only expect an speed increment but not a wow change.

Can you just split the server in two smaller boxes so you have DB AND webserver ? that will do all the debugging process much more easier.

for apache, are you using forking ? if so try tweaking the child processes number, that will give you more concurrency ( the default number of 150 processes is too low, at least on centos)

I don't know much about drupal. Are the system tables InnoDB? I soo try to read a little about how to increase the buffer memory for mysql. Stick to this

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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