I am running Centos 6.5, 64 bit version of Linux. Now I have been getting an error about
my hard drive failure is immenent.

The trouble is I do not know whether to take it seriously. It has been saying pretty much the same thins for a long time - over a year.

I remember running across an error on Red Hat stating that this was a flaw in Red Hat (and hence Centos) ands hould be ignored. What is the story here?

I obviously do not want my hard drive to fail, but I also do not want to replace it needlessly.

Any help appreciated.

Thanks in advance.

R,

jyunker


The error I find on this can be found here:

https://bugzilla.redhat.com/show_bug.cgi?id=498115

What error? Where are you seeing it? Are you running any disk checks or is it just popping up on the terminal?

1. Backup your data immediately (if you haven't done already)!
2. Download the disk manufacturer's diagnosis tool and test the disk for errors.

Just as a note,
I'm running Centos 6.5 64bit on an aged drive and have had no such warnings. Have you checked dmesg for signs its struggling to read the drive? I've seen messages something relating to 'drkby' or such, just before disk demise.

Fred.

There is a gui app call gsmartcontrol, It's designed to check the health status of your hard drive. It works best on hard drives with the the SMART (Self-Monitoring Analysis and Reporting Technology).

PclinuxOS Magazine did an article on using gsmartcontrol

http://pclosmag.com/html/Issues/201401/page01.html

Even tho the article is using pclinuxos as the platform, I'm pretty sure any distro with gsmartcontrol in the repos can follow along.

BTW, the commandline version of gsmartcontrol is called smartctl

I am seeing the error as displayed in the upper right hand corner of the Centos gui. There is a picture of a hard drive with a orange triangle on it. Click on that orange triangle and it says hard drive failure is imminent.

But it has been saying that for a long time (almost a year) and the link in my initial post says this is probably an error.

I did download gsmartcontrol and will try it.

My hard drive is ATA WDC WD3200AAKS -75L9A0.

Any help appreciated.

Thanks.

R,

jyunker

You will probably find that the drive has some small number of reallocated sectors. Some drives store something in the high-order bits of the raw value for that parameter, and the Gnome tools misinterpret that as a huge number of bad sectors.

I have a drive like that (2 bad sectors, interpreted as 32000+). I know that I somehow stopped those tools from trying to read the SMART parameters (Gnome Disk Utility now says that SMART is "not supported"), but I have no idea how I did that and have never been able to turn that feature back on.

1) Backup your data.
2) Run a SMART long test and wait for it to finish: 'smartctl -t long /dev/sda'
3) Post the results and attributes: 'smartctl -a /dev/sda'

Boot with a livecd like systemrescue cd and type "e2fsck /dev/sdX" X=number of failing partition, this will check your partition and tell you if certain inodes are corrupted and might need fixing.

The manufacturer will dis-avow you if they learn that you run Linux.
They cannot cope with Linux, no amount of explanation will do anything. It is not in their book of answers. I wait for the day this might change, but I have not heard of it. (And I have not had a failed drive for a while now.)
Lie to them. Move the drive to a window or DOS something and tell them that is the system
you are using. Do not mention Linux.

The manufacturer disk tool will not likely tell you anything. My experiences with them were that they either see the drive as perfect, or cannot find it.

Use the SMART tools.

A disk drive that has not changed in over a year is rather stable.
My experience is that it goes from multiple bad sectors to a grinding noise in a week, if not sooner. They can get a few bad sectors early, as they are first encountered.
Always backup important stuff to somewhere not on the drive.

Then you've probably been extraordinarily lucky for over a year.

What you're seeing is most likely the S.M.A.R.T. status reported by the drive itself. In that case, obviously the drive can't predict the time of its own failure with perfect accuracy, but there's every reason to take such messages seriously.

You should run the following command and post the results:Replace the device name if your drive is not /dev/sda.