I have a Django Python 256 Slice that has a Nginx front end which serves up static media files, passes Django/Python requests back to an Apache 2.2 prefork MPM server utilizing mod_wsgi. I have two small, low-traffic sites on the slice.

Yesterday I had to restart the slice because the swap was being eaten up. Page load times were 10 seconds plus. Today the memory usage already seams to be spirling out of control. Here are some images of memory usage:

Apache seams to be eating memory like there's no tomorrow. Does anyone have any suggestions to help keep Apache memory usage down?

Here's my apache Config:

Timeout 300
KeepAlive Off
MaxKeepAliveRequests 100
KeepAliveTimeout 15
<IfModule mpm_prefork_module>
    StartServers          5
    MinSpareServers       5
    MaxSpareServers      10
    MaxClients          150
    MaxRequestsPerChild   0
</IfModule>

Edit: changed category.

I know they swap's not being used right now. I said they it was being used yesterday, before I restarted the server.

Currently Apache is using 60%+ memory for two sites that are lucky to get more then 1500 hits a month combind.

How can I make the swap from mpm-prefork to mpm-worker?

I'm getting an error with that install command:

The following NEW packages will be installed:
  apache2-mpm-worker
The following packages will be REMOVED:
  apache2-mpm-prefork
0 packages upgraded, 1 newly installed, 1 to remove and 0 not upgraded.
Need to get 251kB of archives. After unpacking 8192B will be used.
Do you want to continue? [Y/n/?] Y
Writing extended state information... Done
Err http://archive.ubuntu.com hardy-updates/main apache2-mpm-worker 2.2.8-1ubuntu0.3
  404 Not Found [IP: 91.189.88.46 80]
E: Failed to fetch http://archive.ubuntu.com/ubuntu/pool/main/a/apache2/apache2-mpm-worker_2.2.8-1ubuntu0.3_amd64.deb: 404 Not Found [IP: 91.189.88.46 80]
E: Unable to correct for unavailable packages

1. Yes, my apps are still up and working.
2. Both of these commands didn't install anything.
3. Same error as before.

I searched for your issue on google and found a few resources. I would advise you look through them to find your specific problem. I can't vouch for a particular solution as I have not run into this issue myself.

http://www.google.com/search?rlz=1C1GGLS_enUS291US306&sourceid=chrome&ie=UTF-8&q=404+Not+Found+[IP:+91.189.88.46+80]

Glad to hear everything worked out.

As for the sweet spot, it depends on your system resources, the webapps themselves, and the load you expect. Though, the default settings should be more than enough for what you have described. If you want to dial them down feel free to do a little research.

I'm not 100% sure, but I think that using worker for PHP is ill advised unless you are using something like fastcgi to generate your php pages.

Prefork is the default because it is safer. Specifically, it is "more" threadsafe than worker because each request is run in its own process.

Well, my first question is what is the load on that server? If it is low, I would consider lowering your maxclients setting.

consider something like this

StartServers 1
MinSpareServers 1
MaxSpareServers 3
MaxClients 50

With very low settings for start/min/maxservers, Apache with spend a lot of time spawning new ones and then killing them off. Which will probably translate into significantly more CPU usage. You want to find a balance for those settings, so that you aren't running an unnecessary number of Apache threads (consuming memory needlessly) but at the same time you aren't forcing Apache into a very frequent spawn/kill cycle (using a lot of CPU and creating load).

You should also look at the MaxRequestsPerChild settings, and perhaps set it to something other than 0 (unlimited). An Apache process grows in memory size to the size required, but it never shrinks in size. So if a process serves monsterpage.php it will grow to the large memory size required for that page and stay there. MaxRequestsPerChild gives you some control over this. Set it to say 500 and each Apache process will get killed off once it has served 500 requests. So that bloated process that served monsterpage.php will eventually die and Apache will spawn a fresh one, using much less memory, when needed.

Depending on your traffic you might experiment with MaxRequestsPerChild set to 500 - 2000. Generally, a higher setting for a busier Apache server.