WordPress MU

I have Wordpress MU 2.8.4a on Apache 2.1 and PHP 5.2.6. I have site wide plugins Add Meta Tags, Akismet, Popularity Contests, Similar Posts and Simple Tags.

My server is overloading almost within half hour of Apache restarting. Load goes up 100, even 200.

This issue has started happening only recently.

The server maintenance people think that it might be an issue with badly coded PHP. I am not so sure, as it has been running well so far.

Can anyone help? I am willing to make this a paid support request, so if you can help, please email at admin@craftgossip.com. Please send your verifiable credentials within the email.

Thanks!
Vikram Goyal

I just experienced similar issues running on Linode. Moved to another host, but now I get fatal errors our of memory, looks like some type of memory leak.

Where are you hosted?

GoDaddy. So it may be an issue with Wordpress MU you think?

Vikram

What do you mean load goes up to 100 or even 200? !00 or 200 what?

Sorry, I meant the load average on the server goes up to 100-200.

Vikram

Sorry, I'm confused because load on unix boxes is usually 1 for a fully loaded CPU, > 1 for an over loaded CPU and <1 for an underloaded CPU. Its down to how long the CPU queue length is. An overloaded CPU can't process its queue fast enough and you get waiting.

Also how long is this load for? A second? A minute? 10 minutes?

Hi Steve,

The load is there for as long as it takes me or someone else at the server management company to kickstart Apache. The CPU load actually goes upto 100 and even 200.

Vikram

He's asking a 100 or 200 what? What units? Percent? MB?

Hi again,

It's just a load average on a linux server: http://en.wikipedia.org/wiki/Load_(computing).

Typically, as he has said, it should be under 1. Mine is averaging 20+ and reaches 100 - 200.

Vikram

Are you on a dedicated server or shared hosting? Have you any idea what processes are causing the load.

A request for a WPMU back end will fire off a few apache server process (assuming you're using worker and not threaded) and will hit the MySQL server . Unless there is something going seriously wrong which is basically dragging the CPU down its hard to work out what is going on.

Hello again,

I was on a old dedicated server (> 2 years old) and have after all the issues migrated to a new dedicated server in the hope that more grunt power can help mitigate some of the issues.

The processes are all Apache processes which fire off to MySQL. Both seem to together consume all the memory and drag the server down and the swap goes crazy (I am talking about the old server).

So far, it has handled itself, but every now and then the load will shoot off to more than 1. On a powerful Quad Core processor with 4 GB of RAM, with limited traffic of 10000 visitors a day, I would have thought that that wouldn't happen. So I am still worried. What else can I provide to help you figure this out? I am willing to pay and give access to the new and old servers.

Vikram

OK, this might sound silly but are you seeing anything odd in your php error file.

I have had a situation where I got a hanging system which was running a whole pile of apache processes all of which were hanging and trying to access the database.

In the php error log I saw lines that contained the following phrase

in Unknown on line 0

No nothing like that.

Me and my system admin team are quite convinced that it is some sort of loop within MU (either a bug in the MU code OR an inadvertently inserted one by one of the writers of the blogs). We just can't seem to track it down.

Vikram

Whats the maximum script execution time set to?

5000 seconds.

EEK!!

Try knocking that down to 30, ensure error logging is turned on and logging to a file, restart apache and then see what happens.

If you've got a rogue script in there its no wonder its killing things dead.

Thanks for that. That is a valid point, and I can't remember why this was set so high. Will change that to see how it goes.

The error logging is already on.

Thanks again.
Vikram

If it still does it try taking the max time down until it blows something into the error log ;-)

Ahh. Yes. Good points to help debug. Of course, it could be something else altogether. :( I just hope I can track it down

Thanks for your help.
Vikram

I suspect you've got some bad code in there and its just hanging the apache process because of that execution time limit.

Thanks. While I have your attention, could I pick your brains regarding missing images on the new server? After migration to the new server (I migrated by taring the wpmu folder and doing a mysqldump of the database and moving that across). But after the migration none of the uploaded images seem to work. They are all there, but just not showing up. I suspect that it might have something to do with the mod_rewrite issue, but after 48 hours of no sleep, my mind is going numb :).

You can see the issue here: http://craftgossip.com.

Thanks
Vikram

Have you checked the ownership of the image files and folders?

Yes, the permissions and ownership is the same as before.

Vikram