Perc6/i RAID controller - Consistency Check?

[MobileTechie]

Got a firm with a Dell Poweredge with the PERC6/i RAID adaptor. They had a crash with it not rebooting as couldn't find a bootable volume. Checking the raid controller I could see 1 disk of the 3 was offline. Took it out and ran Seatools over it and it failed the test so bad sectors at least.

Got a replacement drive and designated it hot spare and rebooted but no rebuild happened or was an option. The controller seemed to think this was RAID0 but we're pretty sure that it was RAID5. No way of changing the RAID level in the basic GUI.

Anyway, they've got SBS's native overnight backup on USB disk so deleted and recreated the Virtual Disk as a RAID 5 and rebooted. Going into the the raid controller utility the virtual disk now has a bar underneath it showing 3%.and I'm trying to figure out if it's initialising the new raid or not.

The reason I ask is that when I select this VD and select Initialization I get the option to Start Init or Fast Init but Stop Init is greyed out which would suggest it's not initializing. And when I selected Consistency Check I get the option to Stop but not Start that - which suggests it's doing a Consistency check.

So is it trying to rebuild the original RAID5 (assuming that's what it was - very likely I believe) or is this normal behaviour for a new VD?

 

[lcoughey]

I trust that everything is 100% backed up, as there is no undo function if you mess up the RAID. RAID 5 will not stop working with only 1 drive failure. In my experience with getting Dell Perc RAID recoveries in for recovery often, the odds are that there are at least 2 bad drives in the array. Sadly, one of the most common reasons for RAID recoveries to be unrecoverable is because of the damage caused by a tech or end user trying to blindly fight to get it back online with damaged, out of sync drives.

The best approach to this, as I see it, would be as follows:

1. Power down the system
2. Get a full sector-by-sector clone of each drive with ddrescue, RapidSpar, DeepSpar Disk Imager or some other data recovery imager
3. If you run into issue cloning the drives, it is best to seek professional data recovery assistance (assuming that the data is not backed up)
4. If you can get the drives cloned, take a look at the RAID meta data (assuming that it hasn't already been altered) and determine the RAID config and the order of failure
5. Use the clones to virtually reconstruct the RAID with the last two drives that were active before the failure and confirm that you can recover all the critical data (intact and current)
6. Only if the original drives come back as healthy and you are 100% sure you have a backup plan to recover the data should you attempt to fix the RAID

In theory, this should be no more than a 2 or 3 day data recovery project at a professional data recovery lab.

Good luck!

Edit: If you are looking for a professional data recovery lab in the UK that is both trustworthy and affordable, contact Sean at www.pcimage.co.uk

 

[MobileTechie]

We don't need data recovery. As mentioned, I deleted the RAID5 array assuming it would wipe the drives on the basis of having several SBS 2008 bare metal backups, but the info from the PERC interface seemed to be suggesting a consistency . Anyway, it doesn't matter now because it completed and the disks contained no files so it was initialising rather than checking.

The weird thing is the SBS backup restore then failed on the basis that the drive wasn't big enough which implies that it wasn't a single RAID5 afterall. So I stuck a 2TB drive in and it got past that error but errored out with "missing element" meaning it's expecting two drives. Given that, and the fact that it contains 3x 500GB drives, the best guess I can come to is that it had two virtual disks - one a 2 disk RAID0 array (thus bigger than 500GB) and a single disk RAID0 at 500GB

By creating 2 VDs, the first being a new 2TB drive and the second being one of the old 500GB ones, it restores. It wouldn't restore with the first VD being the 500GB drive.

So now I'm going to replace the setup with 2VD consisting of 2 RAID0's and probably move away from the native backup entire given how finicky it's been., maybe ShadowProtect.

 

[SAFCasper]

So now I'm going to replace the setup with 2VD consisting of 2 RAID0's and probably move away from the native backup entire given how finicky it's been., maybe ShadowProtect.

Using RAID0 is exactly how they ended up in the current situation... one disk failure causes the entire server to go offline and requires a restore from backup. Potentially a full days downtime or more depending on speed of your backup media.

Compare this to a single disk failure in RAID1 or RAID10... server keeps running, maybe with reduced performance. You turn up at your own leisure and swap over the faulty drive with a new one. With hot-swap bays this takes all of 20 seconds.

+1 for ShadowProtect. Great software.

 

[fencepost]

Anyway, they've got SBS's native overnight backup on USB disk so deleted and recreated the Virtual Disk as a RAID 5 and rebooted

The weird thing is the SBS backup restore then failed on the basis that the drive wasn't big enough which implies that it wasn't a single RAID5 afterall

This just makes me twitch.

I don't think I'd ever do anything to wipe existing media (working or not) until I knew that the backup was able to be restored on new media. I've seen enough cases where they thought there was a backup but there really wasn't (e.g. tape drive had failed, but they never got reports just kept inserting tapes then ejecting them with the button the next morning).

 

[MobileTechie]

Using RAID0 is exactly how they ended up in the current situation... one disk failure causes the entire server to go offline and requires a restore from backup. Potentially a full days downtime or more depending on speed of your backup media.

Compare this to a single disk failure in RAID1 or RAID10... server keeps running, maybe with reduced performance. You turn up at your own leisure and swap over the faulty drive with a new one. With hot-swap bays this takes all of 20 seconds.

+1 for ShadowProtect. Great software.

I meant raid1's lol

Wouldnt go near raid0 for a server