SA 523 RAID 5 Recovery

[NewportNSW]

I've been asked to recover a server drive failure on an ML 350 G3 with Win2K Server. It's an old unit, with 3 x 36.4G HS drives atached to a Smart Array 523, RAID 5, one logical drive (C: I suppose). One drive had failed.

I hot-swapped an identical replacement drive, the drive activity light flashed for a while, lights went out indicating it's set up as a spare.

After a reboot the lights showed all was good, SA 523 initialisation gave the prompt to turn on Auto Data Recovery (F1). Did that, the light on the new drive flickered indicating rebuilding.

The new drive's lights went out, nothing on the monitor (apart from NTLDR not found, from earlier). I rebooted, the SA gave a message that data recovery was interrupted, press F1, which I did. It went through again, comes up with no online light on the new drive.

A reboot and F2 to bypass recovery and into ORCA utility shows the single logical drive, 3 physical drives and STATUS: Needs Recovery.

I have no documentation on this server (long gone). Naturally the backups are old, not complete, no CD's etc. This server has been moved a few times so preserving the RAID 5 redundant data is the priorty.

I'd appreciate any advice on what to do next.

[Michael A. McKenney]

You might need to install the firmware on the drive and controller so the firmware on the drives match. You could have a bad drive. I suggest rebooting. Starting the rebuild and leaving it alone till all the lights are properly lit and the ACU states healthy. I would give it 2-3 hours at least for 36GB.

[NewportNSW]

I rebooted a couple of times, got an error message after I posted to say the rebuid had failed and run it again. It would run for about 45 minutes maybe. Not long eough I agree.

Thanks for the tip on the firmware. It's apparently an old server, very likely a problem.

[Michael A. McKenney]

I would still get the OS off of RAID 5 and on to RAID 1 or 10. RAID 5 crashes can take out the OS. It corrupts the registry and paging file.

I would not reboot the server once the rebuild starts for at least 6-8 hours. No backups are a major problem. You might need to send the drives to Ontrack. It can cost $2000+ to recover RAID 5. A price of a tape drive.

[gregersenj]

You did not read the information carefully.

F1 = disable logical drive
F2 = Interim recovery mode (Default)

You should have kept your hands away from the keyboard.

At this point. If the logical drive is still disabled, you will have to enable it manually. Press F2.
The on-line LED should begin flashing within a few seconds (Middle LED), and the server should boot strait away.

Why did you reboot the server anyway?

Fw upgrade will do no harm.

BR
/jag

[NewportNSW]

Thanks for the feedback, gentlefolk.

The logical array went offline at failure so I had to do exactly as you said greersenj - F2, then F1 per messages. That got it to the F1 rebuild option.

I had to reboot each time to get it to the next step as it would then try to load an OS and fail. However, as I mentioned earlier, the rebuild didn't go very far each time.

The techs who manage the other sites for this client have stepped in (after 3 years of not managing it).

I think firmware is a good idea, this system is old and original for sure, so I'll suggest it to them.

[NewportNSW]

Update for those who may read this:

We could not get the array bootable, didn't care as all we need is the data. The server is dated 2001, time to go.

Firmware update loaded, SA drivers on a diskette and a variation BartsPE got the data onto a spare IDE drive plugged into the spare IDE port on the CD line. Everything being copied as I write.

This was hard work - getting the array and a spare drive recognised by something that could recover the partition and files therein.

And of couse RAID 5 is a no-no for an OS LD, but I am sure you knew that.

The boot options on HP servers new and old can be painful in situations like this when you want a bootable CD program to recognize a spare drive.

We're calling this done now. If the client knew how you helped, you'd be thanked. PLease accept mine.

[NewportNSW]

Firmware upgrade, know how smart arrays recover and the problems when they don't.

SCSI drivers on diskette. Spare drive and bootable CD with partition and file recovery utilities.