Load Inspection

[jaymc]

wen their are a lot of users on my site, the pages start taking ages to load, even though the load is about 1.9 - 2.3

i was just wondering how i could find out what needs upgrading on my server to help, such as a faster HDD, or faster / Dual CPU's

any linux commands that are good for this....

 

[chirpy]

When this happens, if you could post the top lines or so of top it would help in determining where the bottlneck might be. It would also help to know your OS and kernel release that you're running.

 

[jaymc]

This is at a fairly busy time

18:31:37 up 24 days, 7 min, 1 user, load average: 2.56, 2.30, 2.05
151 processes: 127 sleeping, 5 running, 19 zombie, 0 stopped
CPU states: cpu user nice system irq softirq iowait idle
total 26.7% 0.0% 12.1% 0.6% 1.2% 0.6% 58.5%
Mem: 1554928k av, 1298568k used, 256360k free, 0k shrd, 99072k buff
927780k active, 318780k inactive
Swap: 2040244k av, 684k used, 2039560k free 934484k cached

PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND
28740 mysql 16 0 272M 79M 2072 S 1.9 5.2 19:40 0 mysqld
20539 mysql 17 0 272M 79M 2072 S 1.2 5.2 3:22 0 mysqld
5210 root 17 0 6156 1244 932 R 1.2 0.0 0:00 0 top
5239 nobody 25 0 0 0 0 Z 1.2 0.0 0:00 0 httpd <defunct>
5248 nobody 22 0 0 0 0 Z 1.2 0.0 0:00 0 httpd <defunct>
5256 nobody 25 0 0 0 0 Z 1.2 0.0 0:00 0 httpd <defunct>
13406 mysql 16 0 272M 79M 2072 S 0.6 5.2 17:10 0 mysqld
20623 mysql 16 0 272M 79M 2072 S 0.6 5.2 3:33 0 mysqld
20628 mysql 16 0 272M 79M 2072 S 0.6 5.2 3:44 0 mysqld
942 root 25 0 17656 8024 4044 R 0.6 0.5 10:31 0 httpd
5085 nobody 17 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
5096 nobody 17 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
5241 nobody 25 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
5242 nobody 25 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
5243 nobody 25 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
5246 nobody 20 0 0 0 0 Z 0.6 0.0 0:00 0 httpd <defunct>
1 root 15 0 1796 576 492 S 0.0 0.0 0:04 0 init
2 root 34 19 0 0 0 RWN 0.0 0.0 0:14 0 ksoftirqd/0
3 root 5 -10 0 0 0 SW< 0.0 0.0 0:00 0 events/0
4 root 5 -10 0 0 0 SW< 0.0 0.0 0:00 0 khelper
5 root 15 -10 0 0 0 SW< 0.0 0.0 0:00 0 kacpid
27 root 5 -10 0 0 0 SW< 0.0 0.0 0:00 0 kblockd/0
38 root 15 0 0 0 0 SW 0.0 0.0 1:26 0 pdflush
40 root 11 -10 0 0 0 SW< 0.0 0.0 0:00 0 aio/0
28 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 khubd
39 root 15 0 0 0 0 SW 0.0 0.0 1:18 0 kswapd0
113 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 kseriod
177 root 5 -10 0 0 0 SW< 0.0 0.0 0:00 0 ata/0
179 root 21 0 0 0 0 SW 0.0 0.0 0:00 0 scsi_eh_0
180 root 22 0 0 0 0 SW 0.0 0.0 0:00 0 scsi_eh_1
189 root 19 0 0 0 0 SW 0.0 0.0 0:00 0 scsi_eh_2
190 root 20 0 0 0 0 SW 0.0 0.0 0:00 0 scsi_eh_3
195 root 15 0 0 0 0 SW 0.0 0.0 4:07 0 kjournald

My solution is to build a new server with 4GB of RAM as apposed to 1.5, and have dual CPU's with HT technology. But i thought perhaps its not the CPU or ram..... maybe in need of a SCSI 15000RPM disc or somethin

Any way theirs the TOP any help from that ?

 

[chirpy]

That's a little odd as the Memory usage looks OK and the IOWAIT seems low and CPU while busy isn't excessively so. Again it would help to know the OS and kernel version.

I notice that MySQL processes look very active and you might want to look at tuning them. A query cache can often do wonders. One thing to try might be to use one of the preconfigured ones such as:

cd /etc
cp -f my.cnf my.cnf.old
cp -f /usr/share/mysql/my-large.cnf my.cnf
/etc/init.d/mysql restart

See if it helps.

 

[jaymc]

i have actually tuned my MYSQL conf in accordance with memory spec etc where as you set 32MB for every 512MB of RAM or somethin

any way it is tuned....

one thing i thought, the server is on a 100Mbit line, if their are so many people online then it is possible that perhaps the que time become recognizeable to get on the wire and starts dealing in seconds... as aposed to mls

what ya think ?

also, any solution to somethin like that

CENT OS 4 - 2.6 Kernel Version

 

[jaymc]

any idea ?