Optimising a linux server

TCotton · 02-21-2012, 01:11 AM

Hi there, my first post. I'm not sure if anybody can help.

I've problem with a crashing server at the moment and until I find the time to look into the scripting issue I need to give myself some breathing space by optimising the server to free up some memory and take some of the load off

These are my settings in the httpd.conf

Quote:

fileetag: None
keepalive: On
keepalivetimeout: 5
maxclients: 25
maxkeepaliverequests: 100
maxrequestsperchild: 1000
maxspareservers: 10
minspareservers: 5
root_options: ExecCGI, FollowSymLinks, IncludesNOEXEC, Indexes, SymLinksIfOwnerMatch
serverlimit: 256
serversignature: Off
servertokens: ProductOnly
sslciphersuite: ALL:!ADH:RC4+RSA:+HIGH:+MEDIUM:-LOW:-SSLv2:-EXP:!kEDH
startservers: 5
timeout: 300
traceenable: Off

Do these look like sane settings?

I grabbed some stats from the file:

Quote:

Server Version: Apache/2.2.22 (Unix) mod_ssl/2.2.22 OpenSSL/0.9.8e-fips-rhel5 mod_bwlimited/1.4
Server Built: Feb 20 2012 07:42:43

Current Time: Monday, 20-Feb-2012 16:51:29 GMT
Restart Time: Monday, 20-Feb-2012 13:56:06 GMT
Parent Server Generation: 1
Server uptime: 2 hours 55 minutes 23 seconds
Total accesses: 7258 - Total Traffic: 49.4 MB
CPU Usage: u.89 s.4 cu45.19 cs0 - .442% CPU load
.69 requests/sec - 4926 B/second - 7.0 kB/request
1 requests currently being processed, 9 idle workers

Some other server stats:

Quote:

Current Memory Usage
total used free shared buffers cached
Mem: 655360 450112 205248 0 0 0
-/+ buffers/cache: 450112 205248
Swap: 0 0 0
Total: 655360 450112 205248

Current Disk Usage
Filesystem Size Used Avail Use% Mounted on
/dev/vzfs 25G 11G 15G 44% /
none 320M 4.0K 320M 1% /dev

unSpawn · 02-21-2012, 11:09 AM

Quote:

Originally Posted by TCotton

I've problem with a crashing server

So what are the actual symptoms?
What services does the machine provide?
What runs in your web stack?
Does it crash all the time since installation or only since some time?
Did you run all your logs through Logwatch to hunt for clues?

TCotton · 02-22-2012, 01:43 AM

Thanks for the response

Can you explain to me how I gain access to Logwatch. The next time it crashes I'll check it for clues

I'm getting spikes like below on the CPU but up to 100%, which I guess is when it cuts out

Code:

ladyyb 	babynews.co.uk 	95.0 	[php] 
gasnon 	babynews.co.uk 	99.0 	/usr/bin/php /home/gasnon/public_html/index.php

zhjim · 02-22-2012, 02:45 AM

I would definetly raise the KeepAliveTimeout Value. 5 is just to low and setting this to a higher value does not hurt most of the times. Especially in your case where you are having 9 idle workers and one active request. (Last line of your second Quote 1 requests currently being processed, 9 idle workers ).

Logwatch is a tool you can find through you package repository or here http://sourceforge.net/projects/logwatch/files/

My bet is that you rather check on the script cause it seems that only sometimes it sky rockets the CPU so I guess there is something wrong inside the script or it could be exploited. To further investigate check your access logs and see what parameters the script is called with. This only holds true if you have a form or similar with GET parameters.