Broken RAID 1/0

nickersonm

Hi, I'm new to this forum. After reading some posts, I can see that there is a lot of RAID expertise here, so I think this would probably be a good place to present a problem I have just encountered.

For over a year, I've had a nicely working RAID 1/0 (128k stripes) running on a RocketRAID 404. It was 4 80GB WD's, each on one of the channel Masters (1-4). But yesterday, after a BSOD, something odd happened.

On startup, after scanning devices, the HPT374 BIOS reports a "Broken RAID 1/0", with remaining members being the channels 1 and 2 masters, which is half my drive. There is no option to rebuild, merely Power Off, Destroy, or Continue. If I continue, there is another broken RAID 1/0, looking identical, but this time with the masters of channels 3 and 4. If I continue booting, neither registers as a valid drive. So it seems that my single RAID 1/0 has spontaneously broken into two separate 1/0 arrays, neither of which work.

If I enter the HPT374 BIOS Setup, all the HDDs I have connected are there and visible, and seem to be working fine. The array list shows two identical 1/0 arrays, both with the same name ("RAID10_0"). It also shows the JBOD array I'm running on the secondary channels (made of old HDDs), and the drive list shows the 200GB drive I'm using in single mode. The logical device list only shows my JBOD and single drive, neither of the 1/0 arrays.

I'm running WinXP Pro, and have not installed any new hardware or software recently. I was running on the HPT374 Driver and BIOS version 3.03, but had just updated the drivers to 3.04. I did not restart immediately, and before I did restart, the BSOD occurred. On startup, I went into the HPT374 BIOS settings to upgrade to 3.04, but before I got there the BIOS gave me the broken array error. I shut down for 1.5 hrs. or so, to let my HDDs cool, thinking maybe I had overheated one or two. When I started up again, though, the same error occurred. This time I continued through the errors, and booted off a floppy, and upgraded the BIOS to 3.04 (I do have the 3.03 pre-flash save on the floppy). Nothing changed after I restarted, except the version number and maybe a few colors. I then shut down and re-seated the card and IDE connections, but again nothing changed.

I'm not really sure how to continue - is there some way to merge the two broken arrays into a single, working, array? I haven't tried destroying and recreating the array yet, for fear of losing the data. Just getting 2 drives working in a RAID 0 configuration would be enough, then I could back it up.

Also, now that this has happened, I'm thinking of just running a RAID 0 and doing nightly backups. Is there such thing as a ghost utility that will run from windows, so I can schedule it to run while I'm asleep, without having to manually start the process?

Thanks for your help!

- nickersonm

Tex

I don't know that exact version but often the bios and drivers need to match. Do you need to upgrade the bios on the controller? I have had enough bios/driver upgrades with the hpt bite me over the years I would never risk a raid-0/1 without backing up. 90+ percent of the time things go well IF YOU EXACTLY follow the directions as they vary on if you should upgrade the drivers or bios and exactly the steps to follow. You can even unplug all the drives from the controller at times and flash it and it will see the array again. In your case it seems like its trashed one side of the mirror. Is it seeing all four drives and saying the array is broken or what exactly is the message? I would also email or call HPT tech support ASAP if that data is critical. My advice in general on controllers like this is not to run raid-0/1. Run one raid-0 if you must and BACKUP to the other two drives and use them as single drives non raided for safety. Raid of any kind even raid-1 does not mean you can not backup. It protects against ONE and only one type of failure and thats a mechanical drive failure. Probs with data corruption, virus's, deleting data, flaky controller/driver etc... all eat the whole thing in one big bite. I used to have software to rebuild a raid-0 array from hpt you might contact tech support and see what they can cough up as long as the drive shows and just has a broken array. Don't jack with it at all untill you talk to HPT or you increase the chance of flushing the whole thing if you lose the front raid-0 array. On a higher end controller you would usually be able to break the mirror but keep the first raid-0 with your data but really the HPT/Promise/SI are really toys in comparison and lack the robust features.

Tex

nickersonm

Well, I had not yet restarted to apply the driver update when I got the BSOD. So they shouldn't have been applied yet. When I got the BSOD I was going to update the HPT BIOS before loading windows, but ran into the error. I restarted again and continued through the errors, and updated the HPT BIOS to 3.04, the same version that I had updated the drivers to.

It is seeing all four drives, but seems to have split my single 1/0 array into two 1/0 arrays, with 2 drives each. It gives a "Broken RAID 1/0" error for both of them. I have emailed Highpoint, but I posted here in case Highpoint is unable to solve my problems.

And yea, I should be backing up, but I leave this computer off at night and am generally too lazy to spend time running a backup. Now that this has happened, though, I'll probably start doing nightly backups.

- nickersonm

nickersonm

Highpoint got this cleared up with a utility of theirs to rebuild the array. They certainly have excellent tech support - they got me fixed the first buisness day!

Tex, thanks for your input. I'm going to start doing backups now.

- nickersonm

Tex

If you wanted to be helpful to others consider donating a copy of that utility to Spinner who is a mod for this section and also handles the downloads. That way others with probs can get the utility from S-M's downbloads also.

Cheers
Tex

nickersonm

Sure. I've attached the files here.

hptdet is the detection program, it detects the configuration you have and creates .txt and .bin files with the information in it.

hptrm3 is the array creation utility to remake arrays, for bios version 3.x.

Both of these utilities run direct from DOS, I don't know if they're compatible with windows, but I would guess not.

- nickersonm

Preacher

This is great info. I wished I known about these utilities when my RAID 5 crashed. I luckily had backups, but deleting, rebuilding the array, and copying all of my data was a pain in the butt. Thanks for sharing the files.