? high server load + sites being unaccessible so many times in the day
 

I have a server with 8GIG of ram and a quad-core q9550 CPU and 500GIG of hard-disk space, and 100Mbps net connection.
I have separated it into 3 parts of 2 Gigs of ram each, by virtualization, so there are 3 VPSes.

The problem is that the one of the VPSes (created) with centos and WHM/cpanel would go down very often, as many as 60 times per day!

I asked one of my friends to have a look at it, and he told me that Tomcat was the reason.

Things got a little better (ask me what I changed about Tomcat?!) and after that the server would go down only about 20 to 30 times per day.

I asked the friend again, and he told me that the CPanel integrated antivirus would cause the high load periodically.

(Ask me again, what did I change?) And after that the server would go down only about 4 to 10 times per day.

Now, as far as I can tell, there is nothing more that would/should cause the server to go down.

About 30 minutes ago, before writing this post, the hosted sites were unavailable again, for about 30 seconds.

I checked the load through ssh and I saw that the load is 3, so I went to the main server and checked the VPS ram and noticed that only 10% of the cpu is used and only 307Mb of available RAM is used.
Everything shows that the server should be functioning OK, but it still seems to get unavailable far too often.

I used the top command and noticed that there is nothing using the cpu and the ram and all of them are 0.

I used the mysqladmin proc command and again nothing noticeably weird seemed obvious at this time. I also checked the other VPSes and they were also in relatively low-resource-consumption states

(I don't understand this part):
Also sometimes I use the top command when the load goes high I see that all of the users, even the non visitors users site use cpu seems forexample a refreshing things or updating thing updates all of the accounts forexample mysql or so

Please help me to figure out what other things might be causing these service/server high loads and unavailability.

Thanks in advance!

(NOTE: This post was edited by GrapefruiTgirl in an attempt to clarify some grammar/spelling to make it more understandable)

What is the VM software here? Virtuozzo? Xen?

it is
VMware Infrastructure Web Access

Haven't used that one, sorry.

I think it is
Virtuozzo

What is mean "go down" ?

Not reachable by any service (ssh, www, mysql, ftp ) ?

the main server is rechable

but the vps on it wont be accessible by ssh http
and after 1 min or 2 that the load become alitle lower as I login to ssh I would see that the load is 17 or so
the last time was 10 miniuts ago that this happened
in the vmware I saw that this vps inspite of having high load , the cpu usage was 20% and the ram was only 700MB of 2 gig

First, check if Your system doesnt use too much SWAP.


I suggest to use htop instead of top. Top is old fashion, not readable something...

http://htop.sourceforge.net/


What kind of services You have running on the guest server with centos ?

What kind of service You have running on the host ?

If he has 8 gigs of ram and only 3 vps servers he probably isn't using any swap.

I have installed the htop yesterday
apche mysql and what needs to be run on a hosting server with WHM cpanel

the load is in normal state at the moment
it is 0.52

but in a time it go to mor than 10 and not accessible

Run this and check the failed column:

[root@nd11225 ~]# cat /proc/user_beancounters
cat: /proc/user_beancounters: No such file or directory
[root@nd11225 ~]#

If you don't have that file, it must not be Virtuozzo.

Put this script to Your root directory in the VPS Centos.

http://amos.freeshell.org/userbeans.sh

next execute it using this command (root):

userbeans.sh +a


and please past here result of that script.


Dont worry it will show only reserved resources for CentOS.

root@linux [~]# wget http://amos.freeshell.org/userbeans.sh
--09:39:44-- http://amos.freeshell.org/userbeans.sh
Resolving amos.freeshell.org... 192.94.73.30
Connecting to amos.freeshell.org|192.94.73.30|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 11681 (11K) [application/x-sh]
Saving to: `userbeans.sh'

100%[==================================================================================================== ================>] 11,681 10.5K/s in 1.1s

09:39:45 (10.5 KB/s) - `userbeans.sh' saved [11681/11681]

root@linux [~]# userbeans.sh +a
-bash: userbeans.sh: command not found

chmod 700 userbeans.sh
./userbeans.sh +a

ehhhhhhhhhh...


ehhhhhhhhhhhhhhhhhh

root@linux [~]# chmod 700 userbeans.sh
root@linux [~]# ./userbeans.sh +a
for non-root users, /proc/user_beancounters required via stdin
root@linux [~]#

for non-root users, /proc/user_beancounters required via stdin

execute this like a root...

root@linux [~]# su -
root@linux [~]# chmod a+x userbeans.sh
root@linux [~]# ./userbeans.sh +a
for non-root users, /proc/user_beancounters required via stdin
root@linux [~]#

I found the problem in line 225.

Try replacing that line with this:

if [ "`whoami`" != root ] ; then


Abefroman, can you tell us more about this script or point
us to some web resources?

What are userbeans?

What program(s) creates userbeans?

How can you use this info for diagnostics? (Although the
answer to this may be obvious after seeing a sample output)

This is kinda a dead thread, but the userbeans are for vps servers only and I think specifically virtuozzo.