Can someone help me with several things?

[Divvy]

Hello guys,

Today I contacted cpanel about performance problems due to server load.
And they said this:

Hello

I killed off the pkgacct process and load has came down a good bit. Another source of the high load appears to be a bunch of PHP processes on top of slow disks, ie:

Before I killed pkgacct:

root@server [~]# dd if=/dev/zero of=test_data.bin oflag=dsync conv=fdatasync *bs=8388608 count=16
16+0 records in
16+0 records out
134217728 bytes (134 MB) copied, 32.5411 seconds, 4.1 MB/s
root@server [~]#

Afterward:

root@server [~]# dd if=/dev/zero of=test_data.bin oflag=dsync conv=fdatasync bs=8388608 count=16
16+0 records in
16+0 records out
134217728 bytes (134 MB) copied, 3.68229 seconds, 36.4 MB/s
root@server [~]#

As you can see, the disks on this server (even when somewhat idle) are slow compared to most normal servers so it's causing the server to become sluggish under moderate amounts of load. The model number of your drive (WD20EURS-63S48Y0) suggest it's a 2TB drive @ 5400RPM so it may not be the fastest in your situation. One thing you could do to perhaps lower the overhead of PHP is perhaps disable SuPHP and run DSO with EAccelerator which will give a nice speedup and reduce IO overhead.

There is a newer kernel version available, I also suggest upgrading to that one in yum to see if that helps fix it.

Can someone help me and teach me how to make what he suggest?
1- disabling suphp and run dso with eaccelerator (don't I need that!?)
2- Update kernel

Thank you guys!

Divvy Reviewed by Divvy on 13th Feb 2012. Can someone help me with several things? Hello guys, Today I contacted cpanel about performance problems due to server load. And they said this: Can someone help me and teach me how to make what he suggest? 1- disabling suphp and run dso with eaccelerator (don't I need that!?) 2- Update kernel Rating: 5

[Loonycgb2]

Do you have a vps or dedi with whm/cpanel?

2. ssh into terminal -> yum update kernel

[Divvy]

thank you mate for your reply.

I have an dedicated server.
Only need to run that command? Nothing more will appear after?

Thank you

[SS-Chris]

Hello there,

In order to change the PHP handler, you need to go into:

Main >> Service Configuration >> Configure PHP and SuExec

In there, you can change what the default handler is. DSO is arguably the fastest, however, it is also the least secure. To add a PHP caching engine, you can rebuild Apache via EasyApache in WHM, and make sure you select eAccelerator in the PHP settings.

The root problem is that you have a 5400rpm drive, so reads and writes are slower, which ties up your servers resources as it waits for reads/writes to complete.

BTW - yum update kernel, will not be of much, if any help with this problem.

Chris

[Divvy]

Thank you for your reply mate!

I think that harddisk is a 7200rpm and not 5400rpm like he said.
How can I really confirm that?

root@server [~]# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: ATA      Model: WDC WD20EURS-63S Rev: 51.0
  Type:   Direct-Access                    ANSI SCSI revision: 05
Host: scsi1 Channel: 00 Id: 00 Lun: 00
  Vendor: ATA      Model: WDC WD5000AADS-5 Rev: 01.0
  Type:   Direct-Access                    ANSI SCSI revision: 05

Hmmm I found this in my easyapache:


Im currently using "Mod SuPHP" with "EAccelerator for PHP" checked... It is recommended?
Because It says "v0.9.6.1 - caches compiled PHP pages ONLY when running in a persistent (DSO/FCGID) manner" and Im not running DSO

[ddlanime]

Change suphp to fcgi in WHM

Main >> Service Configuration >> Configure PHP and SuExec

Then go to easy apache and rebuild your apache with eaccelerator module, you can add ore modules by clicking exhaustive list at bottom.


PS: For a longrun, I suggest asking your server provider to replace your HDD with a 7.2k RPM one

[Divvy]

I contacted my host and they said:

I can replace the harddisk with an 7200RPM but then you need to restore your cpanel backup afterwards.

What this mean?
Do I need to config my server all again? and restore all my account one by one?
Is not possible for they cloning the current harddsk to use in the new one?

EDIT: ok they are cloning the harddisk right now... the worst part is that my server will be offline for 6h max