PHP queue getting very deep

[ffeingol]

We're having an issue where the php wait queue is getting very deep and php requests are not getting processed. I'm sure that it's one web site that has a bad MySQL call or something similar backup up all php but we're having a really hard time tracking it down. We're not running suexec. Any hints / tips on tracking it down?

[PSS]

Shut down each website for 10 minutes and see if processes go down?

[ffeingol]

Thnx, but not really an option :-) The problem is not constant.

[mistwang]

When you are having the problem, you can start to "strace" the lsphp process. It may not help much if it is in the middle a request, but it happen to take another bad request, you should be able to tell which script is causing it.

I don't know if you can do anything from mysql side.

[ffeingol]

I was actually trying to debug it from the MySQL side (figuring that was what was blocking). From within phpmyadmin or mysqladmin you can dump the running queries which helps narrow things down.

The problem with tracing the lsphp process is "which one". When things back up there are 40+ lsphp# processes.