Hello

One of our servers seems to be going down often. I've looked through the web server's log files, and have noticed that today we had quite a few entries along the following lines:

[Tue Jul 25 11:03:59 2006] [error] [client 127.0.0.1] (12)Cannot allocate memory: couldn't create child process: 12: php5
[Tue Jul 25 11:03:59 2006] [error] [client 127.0.0.1] (12)Cannot allocate memory: couldn't spawn child process: /usr/lib/cgi-bin/php5

(The IP is ok, these are localhost call-backs which are expected from our web app.)

This happens for a while, until:

[Tue Jul 25 11:04:41 2006] [error] [client 127.0.0.1] (12)Cannot allocate memory: poll failed waiting for CGI child

Then we find that the web server is down.

Do you happen to know why such log entries would be occurring? Any suggestions on how we could fix this issue?

Thanks.

Hi,

What is the RLimitMEM value set to in your httpd.conf? Is it there at all?

cg

Also, how much memory and swap does your system have? If it runs out, it's just plain out of luck. I'd also look and see if the user apache runs as has any specific limits with ulimit.

cg: No, there's no RLimitMEM in there. Should I add it? What's a good setting?

Matir: This is a virtual server hosted on a relatively "large" box. I'm unsure as to the exact limits, but I believe that its around 128mb RAM, 10GB HDD. I'm not sure about swap, as I think thats handled by the super-server.

So this is generally an out-of-memory problem then? Are there any tweaks I can make to apache/php/mysql to reduce memory usage or the length of time they run for before dieing? I've got KeepAlive in apache set to "off" from recommendations.

neocookie: I'm afraid I'm not too sure on what people think is a wise number to set for the RLimit in your situation.

cg