cuziyq, thank you for the quick reply.
I presume your instructions to "Set the controller to normal SATA mode, then boot from a BartPE disk and copy my data off the D: drive to a safe location." imply that the disks' contents should be visible from a different working XP box if I attach the disks to it, but they are not - as I stated in my question.
Why should the disks be visible from Bart's CD? It would not have the drivers too! (In your instructions you do not say anything about it).
Yet you gave me hope. Because I was in a hurry, I used OnTrack's commecial product and it extracted my data off one of the disks (the other was probably the failed member).
So, it seems (to me) that the individual member WOULD be visible from another XP box or Bart's CD, IF its's directory structures were not messed-up (something that the recovery program was able to deal with).
Thanks again,
---
XC
Main Topics
Browse All Topics
by: cuziyqPosted on 2008-10-14 at 09:56:54ID: 22713259
The problem with ICHx integrated RAID controllers is that A) they DO use proprietary structures to manage the array internally, and B) they do it in a half-assed way that still would allow an operating system to "see" an independent array member as an individual disk, but not work properly with it (the LBA mappings are slightly off).
What probably happened is this: The RAID functionality was turned off so that the controller was set to normal SATA. Normally, this wouldn't be a problem, as the driver would load and straighten it out from within Windows.
However, the user probably ran CHKDSK from an XP CD or some other medium that did not have the Intel ICHx drivers on it. It saw the disks as two independent drives, tried to "fix" one of them, and now the two disks are no longer identical. When you boot XP with the controller configured as normal SATA, the RAID driver loads, sees the disks in the array, tries to do its magic, but fails. This is why the array says VERIFY on it (the driver has no idea which disk is the "good" one).
In a normal RAID-1, the controller sends read requests to both drives in the mirror and accepts the first one to respond while ignoring the second. But since they're now slightly different (but mostly the same), read requests coming from one drive do not line up when a read request gets accepted from the other drive, and that causes XP to puke.
Here's what to do:
Set the controller to normal SATA mode, then boot from a BartPE disk (you can make yourself one on another machine using PE Builder at http://www.nu2.nu/pebuilde
Once that's done, set the controller back to RAID and then reformat/reload the system restoring your data. That should take care of it.
Let me know if that works.