Stargate Logo

->  Stargate daily news 
Problems of recent hardware  
Posted by smurphy on Wednesday, 8 September 2010 - 23:49  GMT
spacer line
Stargate Icon

 It is quite amazing. You buy some new hardware, shiny new disk, CPU, memory, Case, DVD-Writer etc. all in, migrate the old system to that hardware, adapt all scripts for automation, monitoring, remote access and apply the newest patches etc.
You tell yourself, that for the next 5 years there won't be any more hardware related issues showing up. That box then runs for 2.5months, and suddenly you are bombarded with S.M.A.R.T. alerts sent to you, and the full-backup is unable to go through any more.

Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA err
# 1  Extended offline    Completed: read failure       90%      4220         233938
# 2  Extended offline    Completed: read failure       90%      4214         233938
# 3  Short offline       Completed: read failure       90%      4214         233938

18734569.984000] ata1: translated ATA stat/err 0x51/40 to SCSI SK/ASC/ASCQ 0x3/11/04
[18734569.984000] ata1: status=0x51 { DriveReady SeekComplete Error }
[18734569.984000] ata1: error=0x40 { UncorrectableError }
[18734570.008000] sd 0:0:0:0: SCSI error: return code = 0x8000002
[18734570.008000] sda: Current: sense key: Medium Error
[18734570.008000]     Additional sense: Unrecovered read error - auto reallocate failed
[18734570.008000] end_request: I/O error, dev sda, sector 233932
[18756960.812000] ata1: PIO error

You go back to the shop you got the computer from - to tell them that you'd need a new harddrive, as the old one is dying (slowly but steadily). And the only thing the people in the support center of that said shop tell you is:

Staff: Well - bring in the computer, and we'll send it in. It'll take around 4 weeks for a repair.

Even though I explain them - that giving in the hardware is not an option, that's all they stick with (because they don't know of any other way, or no one actually told them, that there is indeed another way).
Back home - I contact the Technical support of MSI, and they told me that it's absolutely no problem for me to fill in a RMA form and specify that only the disk is to be replaced.

What does this tell me ?
First - I'm not likely going to buy any computer in that "Professional" shop again.
Second - the Support staff is barely capable of telling you the difference between a computer and a hard-drive - verdict: useless.

I guess they can only impress people by using IT abreviations, but I doubt they really know what these abreviations stand for.

However - to send the disk back in, I need to first get another disk - which I did in another shop of course.
Anyway - now that this disk is being installed - and this time during formatting I do a read/write badblock test (which takes ages), I'll be sure everything is OK on that one.

Oh - I almost forget. I never liked WD disks, and my dislike has been confirmed once again. The old disk is a WD drive (WDC WD5000AAVS-22G9B1). The new one I got is a Samsung (HD503HI). Since 5 Years I am now using Samsung drives in all my systems (if I have the option to choose), I never had an outage in such short time. Server disks died after around 5 Years of service. But that's OK as I had planned for that (Using Raid 1 where ever possible or a decent backup policy). So - all I have to do is replace the drives.

The disk replacement went not as smooth as I'd hoped. The usual issues with /var/run files on Ubuntu systems stroke again. Until I figured it out - 30Minutes were gone. In fact, I was lucky. Almost the same minute I had sync'ed my old disk to the new one (using rsync) and unmounted it, the system crashed. Guess that was too much stress I inflicted in one time. Anyway - tried rebooting from the broken disk - the kernel-files/modules seem to have gone to Nirvana - means the bad blocks have expanded to places where it was fatal. No recovery possible, had to mount the disk externally to try to see what is going on.  

This page was issued from the Stargate Website:

The URL for this page is: