Too EAProc WaitQ

[Prince]

Gentlemen i'm still yet to know what's causing so many EAProc WaitQ in my lightspeed which cause server load and slow loading sites. 20k unique visitors daily.

I have 2 dedicated servers 1 for webpages, 2 for db

Webpages server 1
1. AMD Quad x 8 CORES 2.0Ghz Opteron 6128
2. 24GB DDR3 SDRAM
3. 2 x SAS RAID 1 (15,000 rpm)

db server 2
1. AMD Quad x 8 CORES 2.0Ghz Opteron 6128
2. 16GB DDR3 SDRAM
3. 2 x SAS RAID 1 (15,000 rpm)


mysql configuration
datadir=/home/mysql
tmpdir=/home/mysqltmp
skip-locking
safe-show-database
old_passwords=1
;long_query_time=2
;log-slow-queries=/var/log/mysql-slow-queries.log
;skip-name-resolve
query_cache_limit=1M
query_cache_size=128M
query_cache_type=1
max_user_connections=100
max_connections=100
thread_cache_size=6M
table_cache=27500
key_buffer=1700M
max_connect_errors=20
max_allowed_packet=128M
join_buffer_size=48M
record_buffer=1M
sort_buffer_size=8M
read_buffer_size=8M
read_rnd_buffer_size=8M
thread_concurrency=16
myisam_sort_buffer_size=64M
local-infile=0
concurrent_insert=2
innodb_file_per_table
innodb_flush_method=O_DIRECT
innodb_buffer_pool_size=3G
innodb_additional_mem_pool_size=100M
innodb_log_file_size = 10M
tmp_table_size = 256M
max_heap_table_size = 256M
wait_timeout = 120
connect_timeout = 10
interactive_timeout = 1200

myisam-recover=BACKUP,FORCE

[mysqldump]
quick
max_allowed_packet = 16M

[mysql]
no-auto-rehash

[isamchk]
key_buffer = 128M
sort_buffer_size = 128M
read_buffer = 2M
write_buffer = 2M

[myisamchk]
key_buffer = 128M
sort_buffer_size = 128M
read_buffer = 2M
write_buffer = 2M


Please also tell me what is eating up my memory in my server 1 for webpages.

Click on the link below for screenshots, to tell me the right values.

Thanks.

 

[Prince]

My host LiquidWeb can't help because they don't support lightspeed.

 

[webizen]

the screenshots are too small. pls send.

also i am assuming it is the peak time that you have many EAProc WaitQ, correct?

 

[Prince]

the screenshots are too small. pls send.

also i am assuming it is the peak time that you have many EAProc WaitQ, correct?

That's correct.

I don't know how to make the screenshot bigger.

Maybe you give your email, i send it to you.

 

[webizen]

You can upload to site like http://tinypic.com/ which would not shrink the size and send us the URL.

 

[Prince]

I have update the screenshot.

 

[webizen]

instances in lsphp5 should be set to 1.

also run 'top' from command line on your DB server during the peak time (when you have many EAProc WaitQ) and paste your screenshot here.

 

[Prince]

instances in lsphp5 should be set to 1.

also run 'top' from command line on your DB server during the peak time (when you have many EAProc WaitQ) and paste your screenshot here.

I have changed instances to 1.

Please give more details on how to run it as i'm a server novice.

Thanks.

 

[webizen]

To run it from command line, you need to open a terminal session and issue following command to log into your server:

ssh root@IP_to_your_DB_server

You can also use ssh client program such as Putty (free), SecureCRT to initiate an ssh session.

Once login, issue 'top' at the OS prompt.

 

[webizen]

also increase the following settings to something bigger than 200 (max connections of lsphp5) in mysql config:

max_user_connections=100
max_connections=100

 

[Prince]

I have increase to 250 each, i will wait tomorrow to see how my site perform, but my host was againt it. Read his comment below...

"As I've stated before, this is not a good idea, you will most likely just
overrun your box with mysql connections. The reason things are running slow
has nothing to do with the # of connections allowed but the time at which it
takes to execute them. I can tell this to litespeed as well if you like.

Regards,
Chris"

 

[webizen]

What application are you running on the web server?

 

[Prince]

Wordpress, SMF forum and a social network.

 

[Prince]

I solved the the problem by increasing innodb_buffer_pool_size=3G to 6G and max_user_connections=100
max_connections=100 to 300 each.

Now sites running faster, but i have to upgrade ram in my host server.

 

[webizen]

Thanks for the update.

Do you have any cache enabled at your application level (such as Wordpress, SMF)?

 

[eva2000]

I solved the the problem by increasing innodb_buffer_pool_size=3G to 6G and max_user_connections=100
max_connections=100 to 300 each.

Now sites running faster, but i have to upgrade ram in my host server.

Prince, I use litespeed vps edition on my wordpress blog as well definitely install wordpress plugin w3 total cache, of course with litespeed in use you can disable w3 total cache's options for http gzip file compression in browser cache.

Any reason you set the small file and mmap cache sizes to such a small value as 300 bytes ?

Total Small File Cache Size (bytes)
Total MMAP Cache Size (bytes)

IIRC default is round 20MB (20M) or 40MB (40M)

 

[Prince]

Thanks for the update.

Do you have any cache enabled at your application level (such as Wordpress, SMF)?

YES,
For wordpress i use "quick cache" for SMF i use memcache.

 

[Prince]

Any reason you set the small file and mmap cache sizes to such a small value as 300 bytes ?

Total Small File Cache Size (bytes)
Total MMAP Cache Size (bytes)

IIRC default is round 20MB (20M) or 40MB (40M)

No reason, i just put in those values randomly.

What values should be there?