Average server load for Plesk 8.3 CentOS?

[tungsten]

Hi all,

I was wondering if anyone would be willing to share their knowledge in regards to average server load. We are running the following:

Intel Xeon-Dempsey 5050 Dual Core x2 (3GHz)
4 GBs DDR2 RAM
CentOS 4.2
Plesk 8.3

eVault setup to run nightly at 4am over a few different domains per night.

We have 80 active domains on the box and we transfer about 3-4 GBs of traffic a day. Some days a few GBs more. Of the 80 domains, less than half manage their email on the server. Of the ones that do manage their email on the server, there are only 4 domains with more than 10 email addresses. Logs don't show any abnormal errors. PHP is NOT logging notices.

We are seeing average load numbers as follows:

Last 1 minute 2.35
Last 5 minutes 2.19
Last 15 minutes 1.97

During peak traffic (business hours in North America) we can hold the average around 3.25-3.5.

Having just moved all 80 domains from a Linux RH9 box that never peaked over 1 with the same sites on it, I'm just wondering if this is a normal load for CentOS / Plesk? If this is not a normal load, is there a known culprit for this type of load? Where should I start digging?

 

[Highland]

Yeah, that does seem high. I have a similar box with RHEL5 and PSA 8.3 and under normal load we tend to run 0.5-1.5 with the occasional spike to 3.0 or so (due usually to a spike in spam and the spam filters processing it)

What does top say is doing the most processing?

 

[tungsten]

Top is showing the following right now. However our load is not yet up to speed for the day.

top - 08:50:06 up 10 days, 18:00, 1 user, load average: 2.14, 1.77, 1.44
Tasks: 181 total, 1 running, 178 sleeping, 0 stopped, 2 zombie
Cpu(s): 3.3% us, 0.9% sy, 0.0% ni, 92.0% id, 3.8% wa, 0.0% hi, 0.0% si
Mem: 4148848k total, 3334272k used, 814576k free, 383392k buffers
Swap: 2096472k total, 3716k used, 2092756k free, 2234796k cached


11273 mysql 15 0 163m 46m 5324 S 13 1.2 3480:38 mysqld
22078 apache 15 0 51492 26m 5268 S 6 0.7 0:13.12 httpd
7273 apache 15 0 50876 24m 3912 S 4 0.6 0:00.76 httpd
483 apache 16 0 51148 26m 5228 S 3 0.7 0:04.69 httpd
6427 apache 15 0 50868 24m 3828 S 2 0.6 0:00.43 httpd
22077 apache 15 0 51184 26m 5608 S 1 0.7 0:11.27 httpd
6423 apache 16 0 51112 26m 5164 S 1 0.6 0:01.38 httpd
4312 root 16 0 10496 7276 1360 S 0 0.2 4:44.36 hald
22085 apache 15 0 53412 26m 5716 S 0 0.7 0:14.09 httpd
8264 root 16 0 2828 1020 764 R 0 0.0 0:00.04 top

 

[atomicturtle]

Mysql is doing most of the work. Do you have query caching turned on?

 

[tungsten]

If /etc/my.cnf is the only location under CentOS/Plesk for MySQL config, then no:

[mysqld]
set-variable=local-infile=0
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
#skip-grant-tables

# Default to using old password format for compatibility with mysql 3.x
# clients (those using the mysqlclient10 compatibility package).
old_passwords=1

[mysql.server]
user=mysql
basedir=/var/lib

[mysqld_safe]
err-log=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid

Should query caching be turned on? I'm thinking yes

 

[tungsten]

Hmm... just polled mysql directly with the following:

mysql> SHOW VARIABLES LIKE '%query_cache%';
+------------------------------+---------+
| Variable_name | Value |
+------------------------------+---------+
| have_query_cache | YES |
| query_cache_limit | 1048576 |
| query_cache_min_res_unit | 4096 |
| query_cache_size | 0 |
| query_cache_type | ON |
| query_cache_wlock_invalidate | OFF |
+------------------------------+---------+
6 rows in set (0.00 sec)

Looks like it is on.

 

[Highland]

Try SHOW PROCESSLIST; in mysql and see what that yields

 

[tungsten]

Hmm.. looks like we might have some poorly transitioned queries in one of our sites. It appears to have 30 or so processes in the list. Likely needs to be massaged with the update to the new server. Once we get that cleaned up we shall see how the load is.

Thanks for the direction all.

/beer

 

[atomicturtle]

| query_cache_size | 0 | <- that means its off.

You need to set this in /etc/my.cnf :
query_cache_size= 32M

 

[tungsten]

Thanks ART. We've made the changes to the my.cnf file and restarted mysql/apache. Since restarting, we are seeing the following:

top - 16:52:09 up 12 days, 2:02, 1 user, load average: 2.82, 2.25, 2.26
Tasks: 182 total, 1 running, 179 sleeping, 0 stopped, 2 zombie
Cpu(s): 12.7% us, 0.2% sy, 0.0% ni, 82.0% id, 5.0% wa, 0.0% hi, 0.0% si
Mem: 4148848k total, 4105272k used, 43576k free, 246764k buffers
Swap: 2096472k total, 3784k used, 2092688k free, 3241160k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
18099 mysql 16 0 170m 33m 4532 S 100 0.8 5:08.84 mysqld
18395 apache 15 0 58704 32m 4068 S 1 0.8 0:00.82 httpd
11469 rogers1 15 0 7196 2920 2004 S 0 0.1 0:19.31 in.proftpd
18705 root 16 0 3800 1024 764 R 0 0.0 0:00.94 top
1 root 16 0 2284 552 468 S 0 0.0 0:15.63 init
2 root RT 0 0 0 0 S 0 0.0 0:02.91 migration/0
3 root 34 19 0 0 0 S 0 0.0 0:00.34 ksoftirqd/0

Interestingly enough, we went through and cleaned up one of the old queries that was copying all kinds of stuff to temp tables while running. Not much change ... /shrug

 

[atomicturtle]

Don't expect it to go away, things will just get faster. At least as far as cachable queries go. If you use a lot of views and joins it might not make a whole lot of difference. You might also want to look into using a dedicated server for mysql if you have some particularly sql intensive applications involved.

 

[breun]

Also take a look at MySQLTuner for some suggestions on what to tune: http://mysqltuner.com/