Problem with load Apache + MySQL

Question

I have tried configuring both Apache and MySQL but I still get very high peaks in load: http://d.pr/aD9L

I'm pretty sure these peaks are not caused by increase in traffic, because no peaks can be found in Google Analytics.

How can I find out what is causing these peaks?

Answer 1

The graph you've shown does not show which processes created the load. If the load is low, I would ask if there are:

• Any cron jobs running at this time e.g. backups etc, that would account for the load
• Any cache clear conditions- i.e. a cache is cleared and things are now being served from the app.

It strikes me that these might be the most likely scenarios.

Also, if you are running on a VM, sometimes weird things like this can happen!

Answer 2

Might be good if you can post your config file w/ your question - just be sure to wipe your ip addresses and domain names of course.

There are a number of reasons httpd load can be high from mis-configuration to denial of service -

Let me fire off a few questions your way:::

1. In your logs - are you seeing anything out of the ordinary?
2. Are you using anything such as mod_security, mod_bandwidth, mod_deflate, mod_evasive?
3. Do you have a firewall setup ? does that firewall report anything out of the ordinary ?

4. run this command when the load is

   high ps -aux|grep HTTP|wc -l

5. run this command when the load is climbing

   netstat -ntu | awk '{print $5}' | cut -d: -f1 | sort | uniq -c | sort -n

6. and run this command

   netstat -lpn|grep :80 |awk ‘{print $5}’|sort

7. If you see a large number of unknown ip's in those lists - block them in iptables via the command line like this:

   iptables -A INPUT -s -j DROP

Answer 3

Just to add to chrim2671's answer, one of the best ways to see what's happening is using 'top' - but (AFAIK) that only sorts processes by CPU or memory usage. And you are saying that your problem is load.

In the case of a webserver, all you'll see in top (or ps) is the httpd process running - if you want more details, you'll need to have a look at the logs or server-status.

You should certainly enalbe mysql slow query logging (if its not already on) and see if you can establish any correlations.

because no peaks can be found in Google Analytics.

Google Analytics is a fantastic tool for usage analysis - but it's not a tool for system diagnostics.

Do you understand what 'load' is? Its the average number of processes which are waiting to be scheduled to be run. It can be caused by high amounts of I/O, as well as cpu / memory / network bottlenecks.

Do you have any reason to believe that the increase in this metric is affecting the service provided to your users? Costing you money? If its not having any adverse effect, is it worth worrying about? If you don't know - then you need to have a think about what and why you are monitoring.