This post is more than 5 years old

9 Posts

33260

March 28th, 2011 12:00

migrating RAID 0 configuration to another server

Hi,

I have a dell PE 2850 with 6X HDDs, they are set up (unfortunately) in a RAID 0 configuration. The other day, one of the HDDs appeared to take a nose dive, I get this "BP drive 5" error and a flashing amber on that drive. The drive shows up in Fail state in the RAID config utility and attempts to rebuild it in that utilitiy immediately say "Error".

As it happens I have another 2850 chassis that was unused, so I moved all the drives over to there. In the RAID utility they all seem to come up ok, green LEDs, "ready" in the drive status screens etc, but the logical drive is not seen at all.

I am very wary to mess with the config here since I'm afraid of blowing away the data on these drives. Can anyone tell me what steps to take in the new server's RAID config utility to restore this logical drive (assuming the drive 5 is not actually failed)?

Let me know if more information is needed. Thanks,

T.

11 Legend

 • 

16.3K Posts

April 11th, 2011 09:00

I think the key - looking forward - is knowing what to do with the mismatch.  If the wrong configuration is used, then yes, everything disappears in a hurry.

The RAID configuration is stored on both the drives and the controller, and a mismatch occurs when the configuration on the drive(s) doesn't match the configuration on the controller.  The only place to view/resolve the mismatch is in CTRL-M, Configure, View/Add.  You almost always want the Disk View, and it needs to be saved on exit or the mismatch will remain.

Good luck.

11 Legend

 • 

16.3K Posts

March 28th, 2011 13:00

You're getting an error when trying to "rebuild" the drive, because it is impossible to "rebuild" a RAID 0.  Once it's dead, it's dead.  However, you can often resurrect it by forcing online the drive that failed.

Seeing now as you have moved them, we'll have to see if the configuration settings on the drives have been wiped out or not.  Best to put them back in the original system (unless you know for a fact that the server has other faulty hardware - but what you have seen is not an indication of faulty hardware other than possibly a bad drive - it is by design).  Hopefully, you will get the same thing as before ... all online and one failed.  If so, then go to Objects, Physical Drives, and Force Online the failed drive.  If you're getting anything different, then you may be limited on what you can try, and what you can try may or may not be successful.

 

11 Legend

 • 

16.3K Posts

March 28th, 2011 14:00

Keeping in mind that sometimes a failed drive is a dead drive and assuming that the drive that "failed" is actually still functional, you can run a Verify Media on that drive.  That will include turning off RAID in the BIOS (you'll see a scary message about data loss - you can ignore it), booting to CTRL-A, Disk Utilties, Verify Media (do NOT format), then changing back to RAID (same scary message again), then attempting.

9 Posts

March 28th, 2011 14:00

Thanks for the reply. I won't be in front of the machine again to try it until tomorrow. In the meantime, assuming I can force the drive online, are there built-in utilities etc to try to repair the drive? (chkdsk type functionality, eg to detect & recover from a bad sector or something?)

Thank you,

T.

9 Posts

March 29th, 2011 10:00

Hm, I also see that the drives in the RAID setup utility are 'mismatched' in that bay 4 (ID 4 in RAID Ch-0) has "ONLIN A00-05" while bay 5 (ID 5) has "ONLIN A00-04".  I don't *think* I reversed these when moving them between chassis but now I'm worried.... Is there some way to check on the disk itself what order they should be in (physically)? Does it even matter?

9 Posts

March 29th, 2011 10:00

I forced the drive online and the LED is no longer flashing etc. However, now when booting I get this message:

"unresolved configuration mismatch between disk(s) and NVRAM on the adapter"

Any suggestions on next steps?

Thanks,

T.

11 Legend

 • 

16.3K Posts

March 29th, 2011 10:00

To resolve the mismatch, you need to go to CTRL-M, Configure, View/Add.  If you go into Disk View and NVRAM View, what do you see (careful not to SAVE either yet)?

The PERC 4 supports drive roaming, so it shouldn't matter, however ... you might try swapping the disks to see if it shows any differently.

11 Legend

 • 

16.3K Posts

March 29th, 2011 11:00

"I forced the drive online and the LED is no longer flashing etc. However, now when booting I get this message:

"unresolved configuration mismatch between disk(s) and NVRAM on the adapter"

Any suggestions on next steps?"

Again:

To resolve the mismatch, you need to go to CTRL-M, Configure, View/Add. If you go into Disk View and NVRAM View, what do you see (careful not to SAVE either yet)?

 

11 Legend

 • 

16.3K Posts

March 31st, 2011 09:00

You said you forced online the drive, but which configuration did you use - NVRAM or Disk?  In order for this to work, you have to first resolve the mismatch, so the controller knows which configuration to use.  Choosing the wrong one could result in a corrupt array once you force online the drive.  Have a backup (I hope)?

9 Posts

March 31st, 2011 09:00

I swapped the two disks in bays 4/5, it actually shows the same thing (!?), ie, a00-05 in bay 4 and a00-04 in bay 5.

In any case, I went to Configure>View/Add, but I don't see any way to see the NVRAM config, and the disk info is very sparse, not sure what I'm looking for here. I posted a couple pics of what I am seeing:

First screen after configure>view/add

http://employees.org/~tstevens/temp/IMG00038-20110331-0805.jpg

After pressing F2 for drive info:

http://employees.org/~tstevens/temp/IMG00039-20110331-0805.jpg

After pressing F3 for Logical Drives:

http://employees.org/~tstevens/temp/IMG00040-20110331-0805.jpg

When I press F10 on the view/add screen, it just drops back one screen (ie, exits view/add).

Any thoughts?

Thanks,

T.

 

9 Posts

March 31st, 2011 10:00

It didn't ask which to use, I selected force online and it did so. I was not given any option to choose a config. (I did this from the objects>physical drives screen).

I don't have a backup, but might be able to clone these drives in another machine before trying anything potentialy destructive...

 

T.

9 Posts

March 31st, 2011 15:00

I'm wondering if I have a really old controller bios or something - as I said, there is no option to select from disk or NVRAM configuration when I go to view/add configuration.

I did find an option that appears to just force the controller to read the disk configuration on boot ("BIOS configuration autoselection: User/Disk" under the controller). I guess that's my best option.

T.

11 Legend

 • 

16.3K Posts

April 1st, 2011 11:00

How did that go?  I'm not sure you have many options left at this point.

9 Posts

April 11th, 2011 08:00

Well, so much for that. I had hope when, after forcing the drives online & selecting this option, ESX did load up, but the disks were completely hosed.

I had already made peace with losing my data at that point, so - I started over with a new server, this time with 4 disks in a RAID 1E array. Luckily I had an old copy of a key script that I was concerned I'd have to rewrite from scratch. The rest is just tedious busy work.

Anyway, thanks for the replies & suggestions. Lesson learned I guess. :emotion-4:

Tim

9 Posts

April 11th, 2011 09:00

Yep - thanks again.

No Events found!

Top