W2K overwrote RAID ID on a drive - how to re-write?

[Markus Tener]

Running W2K Pro on a Raid 0 Ultra 160 RAID controller, with 5 drives (scsi ID's 0 through 4).
I was using Disk Manager within W2K to work on a completely different striped set on another controller. Disk Manager always shows the 5 raid drives as unknown (which is ok, because it shows the RAID "virtual" C: drive as normal).

I accidentally right-clicked on drive 0 of the raid array and did something - could have been 'remove volume' or something, which set it back to Dynamic - and must have written a signature to the device at that moment. Grrr!

I immediately realised my error but there's no way to put it back to unknown, so I continued to use the PC (without rebooting) for several hours without a problem.

Now that I've rebooted, the raid controller doesn't see drive 0 as part of the raid array. I assume the unique ID that's written to the drive to identify it as part of the raid array has been overwritten by W2K.

The data on the drive is almost certainly ok because I was able to continue using it for hours. I just need to be able to put the signature back on drive 0 - or create a new one. The system was shut down normally and nothing has been written to any of the drives.

I don't know if a Norton-type low-level disc utility will work but it might. I could try using it to look at the 4 good drives to see if there's a signature somewhere that isn't on the drive 0, and change drive 0, but that could take days.

Assuming that >doesn't< work, can anyone suggest a way to add drive 0 back into the raid array?

The raid utility for this card (SCSI_LCU.EXE) doesn't allow any changes to the array (except "Delete array"!). I couldn't find any switches to allow a debug mode, either.

And, although we use a good backup strategy, we cannot afford to lose the day's work on the array. .

[Markus Tener]

I resolved the problem. Here's how:

1. Unplugged the scsi cable from the u160 raid controller and plugged it into a plain old adaptec scsi controller.

2. Enabled the bios on the card so that it maps INT13H, so that Norton's DISKEDIT can see the drives. (Mine is disabled because I boot from the U160 controller)

3. Reboot using a bootable CD-ROM or floppy which contains himem.sys and the appropriate ASPI8U2.SYS driver (or equiv.)

4. Put Norton's DISKEDIT on a floppy and ran it.

5. DISKEDIT sees all 5 disks and four of them had basically nothing in absolute sector 0; one drive (SCSI ID 0) had sector 0 full of data. This becomes my suspect drive.

6. All the other drives had the same data at the same location, namely

Absolute sector 0 offset 0:
000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
010 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

etc. until

1B0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 01
1C0 01 00 42 FE FF FF 3F 00 00 00 E1 66 3D 04 00 00


7. I copied the existing contents of the suspect drive's sector 0 to a file on a floppy, then zero'd out the sector and wrote the above data to the correct place (after turning off safe mode to allow writing to the drive), then I wrote out the data to the hard drive (ie. flush the modification to disk).

8. Unplugged the cable and put it back on the U160 RAID controller, and rebooted.

9. Disabled the adaptec boot bios again as in step 2 above.

10. Rebooted.

Job done. the U160 controller again sees all 5 drives as part of the raid array and the system boots normally.

Epilog (Step 11): Full backup!

d.