Recovering from a RAID crash - which drive failed?

[wombat2010]

The other day I got the dreaded "A disk read error has occurred | Press Ctl + Alt + Del to reboot" message on trying to boot up my computer. Based on the research I did, I believe I have a corrupted boot sector or failed hard drive. I was using Intel's Matrix RAID feature (on newer motherboards with ICH6R or higher) to run both RAID 0 and RAID 1 using only two physical drives.

The OS and programs reside (or resided ) on the RAID 0 volume and my data was all stored on the RAID 1 volume. I found Intel's instructions on how to recover from a crashed drive (http://www.intel.com/support/chipsets/imsm/sb/CS-021015.htm), which I plan to do. The trouble is that step 2 is to replace the failed hard drive--that is expected, but I do not know which drive failed and I do not know how to find out. I tried running both a Linux distro off CD (Mepis) as well as Bart PE (WinXP boot disc), but neither one could run any diagnostics on the drive(s), even though both seemed to "see" a C: drive and an E: drive (I ended up with an optical drive being named D:, in case you're wondering).

Does anyone have any thoughts on 1) how to determine which drive has a problem (both feel like they're spinning when I touch them) or 2) whether it's okay to try replacing one and then, if that doesn't work, try replacing the other without destroying the RAID 1 volume? I believe I have virtually all of my data backed up (and I definitely have ALL of my pictures backed up, in multiple locations--phew!), but I'd still like to recover the data if possible, both to save time and to avoid losing anything at all.

Thanks much! I knew I should have stuck to plain old drives, but I couldn't resist trying Intel's cool new feature when I built my computer.

Stephen

[David C. Buchan]

Hi Stephen,

This is a risky suggestion as I don't have Raid 0/1 (is one mirroring and the other striping?). You may try rebooting with the power uplugged to each drive in turn to see if that shows up anything. Cool features are good but only if they work.

Regards,

David

[wombat2010]

Thank you David.

According to Intel, both drives are simultaneously mirroring and striping--part of drive 1 and part of drive 2 are acting as a RAID 0 volume (striping) and the remainder of drive 1 and remainder of drive 2 are acting as a RAID 1 volume (mirroring).  I do not know how they accomplish it.

I was just able (finally--Seagate's site was doing maintenance) to download a drive checking utility, which I am hoping will identify the problem drive.  If not, I will try your suggestion.  I will have little other choice.  It's a good thing I have backups.  The moral, as Peter points out in his book, is not to rely on RAID alone for backup.

As for cool features are good only if they work, that is true, but the Matrix RAID has worked for a year and a half for me.  The crash is due to simple hard drive failure, which is inevitable (and only my second out of many drives owned in the last ten years--the other did a lot of clicking to warn me, so I tossed it before it actually died).  Other than figuring out which drive crashed (which I am hoping will not turn out to be a big deal once I try the util or unplug a drive), the recovery procedure sounds straightforward and simple from the instructions Intel gives.  If it goes well (I'll report here), then I think it is a decent poor man's RAID solution.  I guess a more sophisticated (and expensive) dedicated hardware RAID controller would have some way of telling me which drive failed (via email or text message, too, no doubt), but having the control as part of the chipset and being able to run RAID 0 and 1 on two drives is pretty nice.

Thanks again for the suggestion, and I will report back how things go.

Stephen

[David C. Buchan]

Stephen,

Since RAID is designed to protect against hard drive failure, it's a wonder it is so difficult to solve.

David

[wombat2010]

I thought I should close out this loop (those who frequent Hert's forum for IDimager may recognize this--given that my only system younger than 6+ years was totally out of commission, I took the unusual step of posting here and there).

I'm happy to report I've solved the problem, but embarassed to say what the problem was. Recently, something caused my BIOS to be reset to all default values. When I checked on the IDE settings, I relied on my memory instead of consulting the manual for my motherboard. Fortunately, before moving on to more drastic measures to try to figure out what was wrong (swapping out components, etc.), I decided to check the settings (with the manual) one last time--I had just flashed the BIOS, too, so I knew everything had reset again.

In short, I had my RAID settings wrong in the BIOS, so the system wasn't even looking for my array. That explains why, with CD-bootable OS's, I wasn't able to see the drives, even though the Maxtor tool said both were perfectly healthy.

Maybe this thread will help someone else avoid pulling their hair out for 5 days.  I certainly hope.  Again I am reminded to keep those backups up to date!

Stephen