load average: 146.65 - Iowait @ 100%

[GypsyMage]

I'm hoping there's some special configs or something I can do to tweak mysql/php to run with vbulletin.. Help me out!

It looks like mysql threads are filling physical memory and spilling into swap, which is killing the IO on the server.

vBulletin 3.0.1
PHP 4.3.3
mysqld Ver 4.0.18-standard for pc-linux on i686

When under high load, there are no PID's with more than 2% cpu.

p4 3.06 w/HT
1 GB memory

Ideas? Special tweaks? I've done some searches on vbulletin tweaking with mysql, but nothing concrete.

-Ken
http://www.stlsr.com/forums

[Floris]

Have you been through the vbulletin options in the admin control panel? There are some options that could have an impact on performance.

I have just visited your site and it has quite some threads and posts, but not enough to throw the server off its stable path. If you could provide your php and mysql variables then maybe George could give you some advice.

[GypsyMage]

Sure thing.. here ya go!

http://gmage.net/phpinfo.php
http://gmage.net/mysqladmin_extended-status.txt
http://gmage.net/mysql_show_variables.txt
http://gmage.net/my.cnf.txt
http://gmage.net/httpd.conf.txt

[Floris]

I'll poke George and see when he has time to have a looksy - his knowledge is much more then mine on these matters.

[George L]

I'm hoping there's some special configs or something I can do to tweak mysql/php to run with vbulletin.. Help me out!

It looks like mysql threads are filling physical memory and spilling into swap, which is killing the IO on the server.

vBulletin 3.0.1
PHP 4.3.3
mysqld Ver 4.0.18-standard for pc-linux on i686

When under high load, there are no PID's with more than 2% cpu.

p4 3.06 w/HT
1 GB memory

Ideas? Special tweaks? I've done some searches on vbulletin tweaking with mysql, but nothing concrete.

-Ken
http://www.stlsr.com/forums

how many vB users online average ? you got a live link to mysqlinfo.php or extended status script at http://vbulletin.com/forum/showthread.php?threadid=3477 ?

I'd try

1. upgrade PHP to 4.3.6
2. change my.cnf to below and restart mysql

[mysqld]
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
max_connections = 500
key_buffer = 16M
myisam_sort_buffer_size = 64M
join_buffer_size = 1M
read_buffer_size = 1M
sort_buffer_size = 2M
table_cache = 1024
thread_cache_size = 64
wait_timeout = 1800
connect_timeout = 10
max_allowed_packet = 16M
max_connect_errors = 10
query_cache_limit = 1M
query_cache_size = 16M
query_cache_type = 1
skip-innodb

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

[mysqld_safe]
err-log=/var/log/mysqld.log
pid-file=/var/lib/mysql/server0.gmage.net.pid
open_files_limit=8192

[mysqldump]
quick
max_allowed_packet = 16M

[myisamchk]
key_buffer = 64M
sort_buffer = 64M
read_buffer = 16M
write_buffer = 16M

[GypsyMage]

I made that cnf change, but I'm not going to upgrade php. I find it hard to believe that every major board running vbulletin is on the latest and greatest php. I'm blaming my sql/apache/machine setup. Not being critical, and I'm not blameing anyone, just that I don't think I've tweaked something the right way.

Is there any way that I can monitor what mysql or the i/o is doing?

My temporary fix is to cron.hourly a restart of mysqld. Works pretty well so far, but I cringe at the loss of data.

[raxafarian]

what kernel are you running. There is an issue with virtual memory with a combo of redhat enterprise, hyperthreading cpu's, and 2.4.21.-9 kernels. Major symptom: high iowait and having to reboot all the time.

the following solved my problem:
enter at SSH:
echo 2 10 20 > /proc/sys/vm/pagecache
echo 100 > /proc/sys/vm/inactive_clean_percent



also add to /etc/rc.d/rc.local
(to take effect at reboot)



is this a servermatrix server?

more info: http://bugzilla.redhat.com/bugzilla/....cgi?id=114553

[GypsyMage]

Give that man a prize!!

[email protected] [~]# uname -a
Linux server0.gmage.net 2.4.21-9.0.1.ELsmp #1 SMP Mon Feb 9 22:26:51 EST 2004 i686 i686 i386 GNU/Linux
[email protected] [~]#

adding those parms to proc right now and crossing my fingers.

[raxafarian]

2.4.21-9.0.3ELsmp is out
Had same problems with iowait and datacenter recommended running same commands.
I've seen some people say it ran better with a value of 30 instead of 100...depends on your type of load. 100 worked for me so I'm not messing with it anymore

[GypsyMage]

2.4.21-9.0.3ELsmp is out
Had same problems with iowait and datacenter recommended running same commands.
I've seen some people say it ran better with a value of 30 instead of 100...depends on your type of load. 100 worked for me so I'm not messing with it anymore

I'll give up the 2% performance decrease for the added reliability.

Server loads are hanging at .03 again.. mmm

And yes, this is a servermatrix account. Are you with them also?

[raxafarian]

yes..well...TP..same thing.

Hope it works for you. I was a little pissed having a new server that had to be rebooted every 2 days... been over 35 days since I entered those commands (except for a kernel upgrade to 9.0.3)

[GypsyMage]

So far so good.. 1+ hours now. I usually had to restart mysql at least every hour. (cron'd it)