I have a Dell T7500 workstation which is running Redhat Enterprise Linux for two years. I recently installed a new SATA hard drive on the DVD rack and connected it to a SATA cable on the mother board. This was the beginning of a series of errors and failures. The first thing I noticed is that the speed for the new drive is very slow. It takes a few seconds to just display the files under a file folder. But it becomes much faster when I type the same command for a second time. Then my machine starts to have a lot of misbehavior. One thing particularly annoying is that it may stop response when wake up from a screen saver. I can still use putty to ssh to the machine from a laptop and I can see that some process is using 100% CPU. It seems the process is related to the wake of the new hard drive.

I also noticed that when I transfer large files to or from the new drive, a lot of system resources are being used. Sometimes it basically occupy the whole system and I can do nothing else.

This morning when I came to work, the machine stopped response again and I had to force it to restart because I have a lot of work to do. The reboot damaged part of the file system on the new drive and fsck cannot recover it automatically. When I'm writing this letter, I'm still running fsck manually to rewrite some of the blocks so that the system can be restarted.

I feel really frustrated after experiencing all these incidents. I searched online but could not find a quick solution. It seems it has sth. to do with the chip set on the mother board. Somebody mentioned an upgrade of kernel may help but it is basically throwing a dart in the dark.

Any advice, solutions? Thx!

Hello,

what is the size of the hard drive? Some large hard drives are using 4096 byte sectors and there can be aligneemnt issues:

http://www.ibm.com/developerworks/li...-sector-disks/

Regards.

Any related error messages in the /var/log/* files?

HI guys,

The new drive is: Western Digital 2TB WD2002FAEX Caviar Black SATA

I looked up online and it seems this drive is not 4096 physical sector size. So it should not be the alignment issue...

Keep looking.

There are a few errors like:

Device: /dev/sdc, FAILED SMART self-check. BACK UP DATA NOW!

But this is after I forced a reboot. It's not related to the weird behavior daily.

To give you a bit more information, this is what I got from a fdisk -l:

Disk /dev/sdc: 2000.3 GB, 2000398934016 bytes
255 heads, 63 sectors/track, 243201 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sdc1 1 243202 1953514583+ 8e Linux LVM

Is there anything wrong in the format, partitions, etc.?

Replace the SATA cable. If it still does it, the drive may be bad, return if under warranty.

1 members found this post helpful.

Unless it turns out to have been caused by a defective SATA cable as H_TeXMeX_H suggested it looks as if the HDD is defective. You could use smartctl -H /dev/sdc to run a health check on the drive.

I had to return a Western Digital SATA drive.
DO NOT ever mention Linux when communicating with Western Digital, it poisons all efforts to get anything from them. They do not support Linux anything and will tell you that instead of any useful information. Once you mention Linux, anything you say about SMART errors will be like talking to a wall.

I returned the SATA drive to the store, and got another from the same pile. It had the same model number, but had a different circuit board and different jumpers. The first drive could not work with my SATA controller, but the second drive had no problems.

Having SMART errors on a new drive is unacceptable. Those tests are totally internal to the drive and do not involve cables nor controller nor partitions nor formatting. This drive has a serious internal problem and is likely re-reading the sector repeatedly trying to get a good read.

On the second try of the same command, your Linux is processing from the disk cache in RAM, not reading the drive.

You could run badblocks on the drive but the warranty will likely run out before it finishes on something that big.
Running badblocks for 10 minutes will give you an idea of how much trouble that drive is having.
>> badblocks -n -v /dev/xxxx

Hi

Did you consider installing the SATA drive (and cable) in another machine ?
How many hard disks are installed in the system ? If more than one
how are they configured ?
Have you tried running a stand alone hard disk diagnostic/benchmark
program against the drive to measure/evaluate its performance ? (*1)

Are you connecting the SATA drive directly to a motherboard SATA port,
or are you using an inline SATA to PATA to motherboard(PATA) type
configuration?

Is the performance issue purely related to the hard disk ?
Or are you looking at a resource conflict problem that is producing
the symptoms of a poor performing hard disk (see *1)

Have you enabled SMART diagnostics on the machine ? Are there any
error reports ?

Just a few thoughts

C.

It was a bad harddisk. I returned it to Western digital to get a new one. Since then it's been running just fine.