Smart Array 5300 disk "Replacement" status

[Van de Vyvere Carlo]

Hi,

whenever i do a critical update or so to a server I take advantage of the RAID1 config. I power off , unplug 1 HD , power on and put the lun to failed.
If the upgrade fails or so I power off , and change the 2 disks (good one in, bad one out) and i have the old system back in 2 minutes.
I now did it with a SA 5300 , choosed 'FAIL the disk' and tested the upgrade. 5 weeks later i need tot do the same test again (lucky for me this is a tesmachine) so I plug in the original disk , FAIL the upgraded disk but ... no os found ... the original disk is put in "REPLACEMENT" status.
The controller doesn't recognize the original config anymore ? When I plug in the updated disk same happens. I lost all ? Can i fix this somehow ? The RAID config is stored on disk , isn't it ?

I need to redo this test because the update failed on the critical production machine (where the trick with switching the disks succeeded, lucky for me). Now i'm scared to do it again on the production ;-)

Greetz and thx in advance.

[gregersenj · Accepted solution]

I'm sure on what you have actually done, or what you mean with fail the disk.

So here's how to do it, and then you figure out the rest your self.

You have got 2 disk's in a RAID 1 (Mirror)
Shut down the server (Power off).
Remove 1 of the disk's, record it's original position.
Repower the server.
The smart array will report 1 disk missing:
Press F1 to disable Array
Press F2 to start up in interim recovery mode (Default)
You should press F2, or just wait, it will default interim recovery mode after 30 secs.

You do all your upgrades, and then if for any reason you want to go back:
Shutdown/power off server.
Remove the disk.
Insert the disk you want to use in it's original position!!!!.

Yes all RAID informations are stored as meta data in the RIS area on all disks.

So if a disk fails and/or are removed, the RAID will expect a new to be insterted in that position. It will not rebuild to any other position. Except it will use a spare drive if assigned and available.

So in your case a disk will expect to have it's partner in that exact position.

So if you put a disk back alone in its partners position, it won't work.
Put it back were it came from.


Here's an idea for your production!
Buy 2 disks.
Shut down / power off
Remove 1 disk.
power on
wait, untill it boot.
Put in 1 of your extra disks.
Let it rebuild
You are now running on a healthy RAID!

Take the other disk, that you have removed.
Put it in your test server, in its original position.
Start up, You don't have to boot, but you do need to get passsed the RAID controller POST, and leave it in interim recovery mode.
Let it rebuild, and you now got a full healthy spare set :)
for the cost of 2 disks.

Do remember to ensure a good backup, before messing with disks - allways !!


On test.
Put the disk in the original position.
power on
If it remains failed.
boot on the smart start cd
open ACU
enable failed LUN


Note if you put in 2 disks, with different time stamps. With server powered off. The latest/newest will be considered the good one, and it will rebuild from the "new" to the "old" disk.

cheers
/jag

[Van de Vyvere Carlo]

Hi , i give u 10 points for :
*confirming the procedure I always use
*putting me into the direction for solving my problem.

I booted from the smartstart CD into ACU but the re-enable option wasn't there. 1disk was failed (in absence) the other was seen as a replacement disk. I tought it was a lost case. BUT , then i plugged in the second disk, rescan and the re-enable option appeared. Plugged out the disk and re-enabled .... tataaaa ... it worked like a charm. i repeated the procedure with the other disk and now both disks are available !!! Thx again !
Carlo.

[gregersenj]

Glad to help :)

thanks for the points.

cheers
/jag