bad blocks on hdd... not bad ???

[Thierry ITTY]

Hello

I've a server (standard intel pIV) with several 200GB disks (5 Western
Digital) for the data and a 60 GB for the system. some disks are connected
to the mobo ide (intel ICH4) and some to an additional ide card (Silicon
Image PCI0680).

I've got the following errors on almost all 200 GB disks (but none on the
system disk hda)

kernel: hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hde: dma_intr: error=0x40 { UncorrectableError }, LBAsect=4194146,
high=0, low=4194146, sector=4194144
kernel: end_request: I/O error, dev 21:00 (hde), sector 4194146

very early. thought, it worked a while (maybe users didn't need the files
with errors ?). the things were worse so I decided to check it all from the
beginning.

I downloaded WD tools, checked all the disks (75 mn low level scan), and no
disk gave any error. then I rebuilt the LVM (one VG, one LV) and the file
system (920GB XFS). everything went fine for 2 days.

now I'm getting those errors again.

WD says it might be a software bug (linux drivers, kernel, bios...) causing
those unreadable sector errors...

I'd really appreciate any clue...

PS I set the machine bios up-to-date during tests, no change.
PS2 and actually I have 2 identical machines with the same problem (so
obviously I can't think of an hardware problem !)
PS3 both machines have high-grade UPS so neither is it an power supply
problem !

TIA