Bad Disk IO on DL380 G4

[Simon Adams_2]

I have a server running SQL Server. It is occasionally reporting IO errors and torn pages.

It has the following disk config:
6i Controller - v2.32 firmware
Two sets of mirrored disks:

Logical Volume 0 RAID1 (OS and App):
72.8Gb, model - COMPAQ BD07288277 - Firmware HPB0
72.8Gb, model - COMPAQ BD07288277 - Firmware HPB0

Logical Volume 1 RAID1 (SQL Data):
72.8Gb, model - COMPAQ BD07288277 - Firmware HPB0
72.8Gb, model - COMPAQ BD07287B4C - Firmware HPB3

The link at http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_EX050119_CW01

appears to describe the problem perfectly for Logical disk 1 - the drive hosting the SQL Server data files - and mentions the EXACT model and firmware versions of disk affected by this issue, ie the Model BD07287B4C disk. (It is also noted that we dont have the same model disks for logical drive 1).

I have run HDDETECT and confusingly is says all the drives are ok. Even if I change all the other drives in the survey.txt files so the model numbers and firmware versions match the know bad versions, it still says 'all disk units are good'.

So the first question is: is HDDETECT working correctly? Or do I really have no problematic disks?

Secondly, the write errors were seen when the 6i cache was set to 50% Read/50% write. Since changing this to 100% read / 0% write, we have not been able to replicate the problem.

We are concerned that we appear to have a disk ID that matches known bad disks and confused that changing the cache settings on the controller 'appear' to have resolved the problem - despite extensive testing in SQL Server.

Can anyone help explain this behaviour?

Kind regards

Simon