Smart array P400 controller RAID drives order important?

[Murad_3]

If a DL380 G5 is shutdown and then has all the RAID drives removed at once, do they all have to go back in the same order they were in originally.
Two logical drives: RAID 1+0 2 drives
RAID 5 = 4 drives.

We had an emergency situation and had to remove drives(v quick) due to flooding. Does the controller manage reisnserting drives in the 'wrong' order well?
There wasn't time to note the arrangement.
And I am waiting to get to the documentation.

Similar situation for an old ML350G3 with a 641 controller.

Also, if you have any lifejackets, could you throw us one?
TIA

[Taifur]

Hi,

You should to reinsert harddisk as before otherwise RAID configure will be misconfigured.

RGds//
Taifur

[Murad_3]

Thanks Taifur.
I will have to wait for the documentation then.

[Murad_3]

Actually-when you say, misconfigured, does that mean there would be any irreparable damage to the array?

Or does it simply not allow the start up?

[Taifur]

Hi,

>>Two logical drives: RAID 1+0 2 drives
>>RAID 5 = 4 drives

If your first 2 drives contain OS and next 4 drive contain data, so you should reinsert accordingly as before, because array controller will not understand where, the os and data was exist before

>>Actually-when you say, misconfigured, does that mean there would be any irreparable damage to the array?

yes, there would be any irreparable damage to the array and may be not allow to start up


Rgds//
Taifur

[Murad_3]

Is there any way of working out what the order is? Any way of identifying each drive one at a time?

[Jimmy Vance]

The really old Smart Array controllers the drives had to be re-installed in the same position. That hasn't been the case for some time now.

With the P400, if you re-install the disks in a different order, you should see a message during post that the drive order has changed and the information is being updated. The same should hold true for the 641 as well.

[Jimmy Vance · Accepted solution]

You can also run the array diagnostic utility

[BenS]

Actually, I'm going to go out on a limb, and say that they do not have to be re-inserted in the server in the same order.

I beleive the RAID information is stored on the disks themselves. The idea being that if you experience a critical server failure, or something similar to what you just experienced, you can swap the disks out of one server, and put them in another, and the the RAID controller should be smart enough to detect changes in the positions of the drives

see here:

http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c01608507/c01608507.pdf

From Page 51:

"You can move drives to other ID positions on the same array controller. You can also move a complete
array from one controller to another, even if the controllers are on different servers.
Before you move drives:
â ¢ The server must be powered down.
â ¢ If moving the drives to a different server, the new server must have enough empty bays to
accommodate all the drives simultaneously.
â ¢ The array must not have failed or missing drives.
â ¢ No spare drive in the array is acting as a replacement for a failed drive.
â ¢ The controller is not running capacity expansion, capacity extension, or RAID or stripe size
migration.
â ¢ The controller is using the latest firmware version (recommended).
Before you move an array to another controller:
â ¢ If the other controller is already connected to one or more arrays of configured logical drives, the total number of logical drives on the controller after the drives have been moved must not exceed the number of logical drives that the controller supports. This number depends on the controller model
and on the controller firmware version.
â ¢ The total number of hard drives on the other controller after the drives have been moved must not
exceed the maximum number of supported hard drives for that controller model and firmware
version.
â ¢ All drives in the array must be moved at the same time.
When all the conditions have been met:
1. Back up all data before removing any drives or changing configuration. This step is required if you
are moving data-containing drives from a controller that does not have a battery-backed cache
module.
2. Power down the system.
3. Move the drives.
4. Power up the system.
o If a 1724 or 1727 POST message appears, drive positions were changed successfully and the
configuration was updated.
- If a 1785 POST message appears:
i. Power down the system immediately to prevent data loss.
ii. Return the drives to their original locations.
iii. Restore the data from backup, if necessary.
5. Verify the new drive configuration by running ORCA or ACU ("Configuring an array" on page 25)."

I just tried jumbling aroung the drive positions on a DL385 G3 we have in our test lab. On bootup, we get an error "1724: Physical Drive Position Change(s) detected - Logical drive configuration has automatically been updated". After that, the machine booted up fine.

Not sure if this would work if the disks were put into a new server with a different RAID controller, but the doco seems to say that it would. (provided you followed the instructions in the doco)

[Murad_3]

Thanks Jimmy and Ben. I know that our mistake was not labelling the drives directly but instead using a chart and printing off the array controller info from the system management homepage as reference. Which may well be floating around somewhere with the rest of the office.

I will try it with the DL380 and I have found some older docs for the ML350 with the 641 controller which do have the order-except I know that at least one drive was replaced since that date so will have to cross reference the drive ID tags with the paperwork.
The recovery guys are checing out the drives now to see if there was any damage-I was loath to try the drives in case there was any residual water etc that we didn't notice in the rush to get out. It was my Poseidon moment. Where is Moses when you need him?

Good thing the most recent full backups worked for both so if worst comes to worst we can always go back to Square 1.0.