Apache2 won't close children & eats memory

[Peter Hase]

Hi Forum,

I looked through the forum here and used google to try to find an answer to my problems:
I recently migrated to a new VPS with my previous plesk installation via the migration manager (Plesk 10.4 to 10.4, sam os, same version: Ubuntu 10.04 LTS).

Now my Apache web server is configured for consuming the least amount of ram and cpu resources, at least I hope to. I monitor it with the plesk health monitor.

My apache2.conf looks like this (in the important parts):

# Timeout: The number of seconds before receives and sends time out.
#
Timeout 10

#
# KeepAlive: Whether or not to allow persistent connections (more than
# one request per connection). Set to "Off" to deactivate.
#
KeepAlive On

#
# MaxKeepAliveRequests: The maximum number of requests to allow
# during a persistent connection. Set to 0 to allow an unlimited amount.
# We recommend you leave this number high, for maximum performance.
#
MaxKeepAliveRequests 500

# KeepAliveTimeout: Number of seconds to wait for the next request from the
# same client on the same connection.
#
KeepAliveTimeout 5

##
## Server-Pool Size Regulation (MPM specific)
##

# prefork MPM
# StartServers: number of server processes to start
# MinSpareServers: minimum number of server processes which are kept spare
# MaxSpareServers: maximum number of server processes which are kept spare
# MaxClients: maximum number of server processes allowed to start
# MaxRequestsPerChild: maximum number of requests a server process serves
<IfModule mpm_prefork_module>
StartServers 1
MinSpareServers 1
MaxSpareServers 7
ServerLimit 30
MaxClients 30
MaxRequestsPerChild 500
</IfModule>

What happens is that the process spawn children, but doesn't close them - the memory just slowly slowly groes. Like a memory leak, but I don't know where it comes from. I tried to minimize the number of mods used and disabled SSL in all my vhosts. I am using stock configured mysql and php5.

Here is what apache2ctl fullstatus tells me when I run it twice shortly after another:

Restart Time: Wednesday, 28-Mar-2012 13:16:07 CEST
Parent Server Generation: 1
Server uptime: 23 minutes 39 seconds
Total accesses: 321 - Total Traffic: 12.4 MB
CPU Usage: u21.6 s2.59 cu0 cs0 - 1.7% CPU load
.226 requests/sec - 9.0 kB/second - 39.7 kB/request
1 requests currently being processed, 6 idle workers

____W_._......................

Current Time: Wednesday, 28-Mar-2012 13:39:59 CEST
Restart Time: Wednesday, 28-Mar-2012 13:16:07 CEST
Parent Server Generation: 1
Server uptime: 23 minutes 51 seconds
Total accesses: 322 - Total Traffic: 12.4 MB
CPU Usage: u21.6 s2.59 cu0 cs0 - 1.69% CPU load
.225 requests/sec - 8.9 kB/second - 39.6 kB/request
1 requests currently being processed, 6 idle workers

___W__._......................

These is what the idle childs are doing (according to apache2ctl and without personal details):

4-1 30117 0/3/12 _ 0.00 11 0 0.0 0.01 0.02 127.0.0.1
<generic vps host> GET /server-status HTTP/1.0
11-0 - 0/0/33 . 13.34 621 0 0.0 0.00 0.30 127.0.0.1
<generic vps host> GET /server-status HTTP/1.0
12-0 - 0/0/26 . 1.39 621 0 0.0 0.00 1.36 127.0.0.1
<generic vps host> GET /server-status HTTP/1.0
13-0 - 0/0/24 . 1.13 621 19 0.0 0.00 1.91 72.14.199.202
<actual vhost configured> GET /feed HTTP/1.1

Does anybody have an idea what is going on here?

 

[Peter Hase]

kinda solved...

I singled out the culprit: It was PHP!
I was fiddling with all kinds of Apache2 settings, but didn't resolve the issue. Then I switched from mod_php to fastcgi php and bam, the problems vanished.
Now apache2 takes around 16MB ram instead of up to 600MB and the php5-cgi children kill themselves as they should after the designated timeout.
The initial loading of a page takes a bit longer, but that is a ok for me.

So, I guess, the problem of the mem leak and and bad php script somewhere isn't solved, but at least switching the php module helped me a lot. for now.