badblocks and hotswap

[aiki2]

Hi all, I have a new intel server with a hotswap kit for 3 sata disk. I have never worked with hotswap before. So, if I run badblocks on the disks with the hotswap kit I have errors BUT running the same test outside the kit, this is connecting the SATA disks directly to the motherboard, I have no errors.
I'm running badblocks with a gentoo 2007.0 live cd.

Any ideas?

[NeddySeagoon]

aiki2,

Its unlikely you have real bad blocks as the drives internal logic maps bad blocks to spare space. Normally when an imminent read failure is predected, so you don't lose any data. This way, the disk always shows the operating system that it has no bad blocks, until the spares are all used, at which time the drive is end of life.

You problems probably lie in the SATA wiring.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[aiki2]

thx NeddySeagoon for your answer, I'm sorry but I don't fully understand it. I run the badblocks command with the -w option so I have a read/write test.

[NeddySeagoon]

aiki2,

The errors do not occur within the drive. When the drive is made some cylinders and some sectors are allocated as spare.
They are not counted in the capacity of the drive, nor are they available to you directly. The firmware on the drive tracks performance of reads and when it predicts a block is about to go bad, it maps one of the spares in place of that failing sector and rewrites you data there. You cannot see this process happening but you can use smartmontools to see how many spare sectors are used.
When all the spares are used the drive cannot hide bad blocks any more and you see genuine bad blocks appear.
When this happens the drive is worthless and should be thrown away.

When you run run badblocks -w, to do a read / write test, you do know that the system receives different data in the read to the write.
Before this bad block remapping was in place (any drive over 4Gb) in the drive, it did indicate real bad blocks.
With modern drives we can be fairly sure that the issue is elsewhere, as the drive hides bad blocks.
This is reinforced by the different outputs depending on where the drive is connected.

In short, you drives are OK and the problem is elsewhere.

I suspect that you get different bad block lists from ever run of badblocks. That too points to something other than the drive.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[aiki2]

well... now I get it!
thanks a lot for this explanation.
Hope find the reason for this behavior.

[aiki2]

well, figured out why, the disks are not compatible with the motherborad (intel s5000psl). So, this strange behavior. I've run some test with disks supported by intel and everything works ok.