[Resolved] High CPU spikes everywhere

[Raenk]

Hi,

I've been having ridiculous high CPU % from some processes according to TOP, but the average load remains below 1, it is quite strange, some examples:

5615% CPU lshttpd
1142% CPU for /usr/bin/mysqld
859% CPU crond

Server: XEN VPS cPanel, it is not swapping or running out of memory in any way and as I said, the load average is low. 3.0 CPU Ghz, 2GB RAM.

The server is hosting only one (joomla) site with an average of 800 visits daily, 6,000 page views average. It has mixed Innodb and MyISAM tables on its DB.

Server has FastCGI compiled, but using LSAPI (SuExec), has SuHosin, eAccelerator, Zend Optimizer, ioncube and mono, although it is not in use.

LiteSpeed is up to date.

At WHM, checking Main >> Server Status >> Daily Process Log
The webpage process (user) takes too much CPU% when I don't think it should for the number of visits it takes.

I have another server, exact specs except for LiteSpeed and Mono, which is running 15 website aprox. and 3 of them together, wordpress based, can put up to 170,000 visits monthly without above CPU spikes, nowhere near it !

What are your thoughts ? If you need more info please let me know.

Thanks

 

[meto]

Do you have LSWS 4.1 and AIO enabled?

 

[Raenk]

I have LiteSpeed 4.0.15
AIO is not enabled, should I ?

Thanks

 

[meto]

Such CPU usage i got on 4.1 rc2 with AIO enabled. Sorry no other clues...

 

[NiteWave]

does it happen frequently ? how about now? 5615% CPU really high.
at the same time, is IOWait high?

 

[Raenk]

It can happen all day, you know, spikes here and there different processes while 'top -c' (for instance, just now, 142% CPU for named process).

At cPanel, for today, the highest is: Top Process %CPU 1736 lsphp5

vmstat command

procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
r b swpd free buff cache si so bi bo in cs us sy id wa st
0 0 0 84600 227484 1506892 0 0 2 4 7 7 0 0 100 0 0

uptime command:

20:10:22 up 19:03, 1 user, load average: 0.10, 0.10, 0.03

I moved eAccelerator cache to /dev/shm and it is writing, but don't see any changes regarding this issue.

free -m command

total used free shared buffers cached
Mem: 2023 1942 80 0 222 1471
-/+ buffers/cache: 248 1775
Swap: 2047 0 2047

shm_size for eAccelerator is 1

It is weird.

Thanks

 

[meto]

In VMstat I can see that you have 100% Idle CPU or am I wrong?

 

[mistwang]

with 5615% CPU, the server load should be around 56, I think it is something wrong with TOP output.

 

[Raenk]

with 5615% CPU, the server load should be around 56, I think it is something wrong with TOP output.

Yes, I think there is something weird going on, also, TOP has started to show that swap is being used, however, 'free' shows zero.

Now get this, it is the new highest level today: Top Process %CPU 9472 lsphp5 while: [~]# uptime - 15:31:39 up 1 day, 14:24, 1 user, load average: 0.06, 0.06, 0.01

That is shown at cPanel Main >> Server Status >> Daily Process Log

It doesn't add up.

 

[Raenk]

Ok, we went deeper on this and found out that it was a timming issue in the kernel, we have patched it and so far it seems fixed.

Thanks for all your replies