Change small disks to bigger disk in RAID

[FerdiO]

We got a DL360 G4p server with 2 x 146GB harddisks (Smart Array 6i Embedded). Fault tolerance = RAID 0+1 (seen from Array Configuration Utility) It's running Windows 2003 Standard SP1 (C is 10GB, D got the rest of storage).

It's being used as a SMS-server (also distribution). We want to replace the existing disks with 2 x 300GB harddisks. What is the best way of doing this without losing data offcourse?

Any help would be appreciated.

PS: We use Dataprotector for backup & restore.

[Víctor Cespón · Accepted solution]

Power off, replace one disk
Let the RAID rebuild
Power off, replace the other disk
Let the RAID rebuild

Then you'll have a logical unit of 137 GB and 142 GB free. You can use that free space to extend the logical unit.

This will NOT modify Windows partitions, as that's has to be done by the operating system.

[FerdiO]

Thanks for the quick response.

I could use diskpart.exe to extend the partition. Perfect. That is much easier then I thought. Didn't know if RAID would accept 2 different harddisks.

[Uwe Zessin]

vcespon,

I respectfully, but STRONGLY disagree with the power-off operation on a SmartArray controller.

I had one colleague overwrite my data of a RAID-1 member, because he replaced the other member (I don't remember the details why) of that disk drive array during power-down. After power-up, the RAID-1 reconstruct too the new disk drive as a source which meant 'good bye' to my data.

[FerdiO]

Then it would be better to following??:

1) power down server
2) remove disk 1
3) power on server
4) place disk 3and let mirrorset build
5) power down server
6) remove disk 2
7) power on server
8) place disk 4 and let mirrotset build

Disk 1 + 2 are small disks
Disk 3 + 4 are big disks.

[Uwe Zessin]

1.) keep power on
2.) pull disk 1
-- will be recognized as a failed disk
3.) wait 1 minute or so
4.) insert disk 3
-- will be recognized as a replacement and
-- RAID-1 reconstruct should start automatically
5.) monitor reconstruct with online-ACU
-- carefull watch until the reconstruct has finished

6.) pull disk 2
-- will be recognized as a failed disk
7.) wait 1 minute or so
8.) insert disk 4
-- will be recognized as a replacement and
-- RAID-1 reconstruct should start automatically
9.) monitor reconstruct with online-ACU

10.) after 2n reconstruct has finished
-- should see that there is free space in the disk drive array - expand (or was that extent? - I never can rememeber) the logical disk
and so on...

[Víctor Cespón]

Id the disks are new, they shoud be recognized as blank disks and the rebuild performed from the disk with data to the blank disk.
Also, you're right, inserting any disk with the server ON, will make it to be treated as a replacement disk, erased and used to rebuild.

[Uwe Zessin]

I agree that 'new' disk drives should not have SmartArray meta-data, but over the years I've learned to be _very careful_ (I've just checked a production server before I put my hands on it - bad fiber cable - good that I noticed it in advance ;-)