Upgrading/Replacing Server Drives

Quint

Quint

Quint

Associate
Joined
22 Feb 2007
Posts
2,137
Location
UK
I need to replace the harddrives in an ML110 G5 for bigger drives.

Its currently configured as Raid 1 (2x 500GB drives) and I need to upgrade/replace them with 2x 1Tb's without reinstalling Windows Small BUsiness Server 2008.


My plan is to use Backup Exec System Recovery 2010, take an image. Wip out the drives and put in the 2x TB's raid 1 again and try and recover the image.

This sound ok?
 
I've never used the backup solution you propose, but that's how I'd do it (Although, probably using a Linux Live CD with dd to dump/restore the disk image).
 
Meh, too complicated!
Pull one 500 drive, swap with a 1tb, let the array rebuild, take out other 500 drive and swap for the 1tb and let that rebuild.(full system backup taken before of course :D)
 
kippermitten said:
Meh, too complicated!
Pull one 500 drive, swap with a 1tb, let the array rebuild, take out other 500 drive and swap for the 1tb and let that rebuild.(full system backup taken before of course :D)
Click to expand...

that would only allow 500Gb of 1Tb HDD array to be used.

Stelly
 
kippermitten said:
Create an array in the unused space!
Click to expand...

If it's the embedded RAID controller on the box, then I'm not sure what expansion options it would support...

An ML110 G5 will take 4 x 3.5" NHP disks, so you might as well avoid the agro of disk pulling and just stuff the 1TBs into the free bays (assuming the box only has the 2 x 500s in at the moment).
 
I am using the embedded controller.

So if I stick in the 2x 1Tb drives on the spare sata slots, build a 2nd array, boot into windows, will that see them in the Disc Managment? Never thought of that, I'd sooner replace them though, I think the current HD's are on their last legs, it'll be more peace of mind.

kippermitten - I think you can only do this if the controller supports it.
 
Given it's the basic embedded controller, I'd go with the extra array or the BE SR 2010 imaging over trying a RAID expansion.

Does your copy of BE SR support P2V? If so, I'd probably take an image, then try to run it up on vmWare to ensure I had an additional recovery point before trying the disk replacement. The worst case fall-back point would to put the pair of 500GBs though you'd need to mess with the RAID controller to re-create the mirror as it would need to reset to reflect the 1TB HDs.
 
This is a mine-field operation.

I did a proof of concept on an image process on test equipment, it worked. I did it live and the OS decided to assign D to the C drive after successfully booting it on new drives for the first time then rebooting after. I lost the ability to boot the system - PERIOD.

DO A FULL BACKUP FIRST. TEST IT. KEEP IT SAFE.

Do a full block level image of the partition WITH VERIFY.
Pull your existing disks, make note of which slots they came from in-case you need to revert back to them.
Put in a new set of disks, you will likely need to nuke the old array that is now missing through the ACU, set up the new array etc etc.
Restore the image to the disks and hopefully you can boot.

After you have done it you will likely need to write a new signature to the disk, Volume shadow copy might well break after you put the new image on a new raid array. The system I worked on had its VSS on the system partition die, writing a new signature was the only solution.

Think very carefully about what media you are going to use to store the image on and from where you will be pulling it when restoring it. Ideally you want to pull it over the network with an agent which supports it, the less disks you have in the server while doing the operation the better, best case is the raid array for the partition you are restoring only.

I've not worked with an embedded HP RAID controller to date that support the advanced expand options you will require to do this any other way. Do any exist?
 
Last edited:
I haven't done a disk expansion on a Manufacturer box, but the box did have a RAID array.

We whipped out the 80GB disks that were in there, ghosted each of them to bigger drives. Nuked the array and recreated it with the new drives. Bob's your uncle, server was up and running again.

Probably would never happen again though haha!
 
Test worked as follows.

-SBS 2008, 2x HD's - Raid 1.
-BESR 2010
-Netgear Readynas (2x500GB HD's- Raid1).

Installed BESR2010, backed-up over the network to the Nasbox.
Pulled the old drives out, installed new HD's, booted into raid, created raid 1.
Rebooted with BESR2010 recovery cd, installed raid driver, mapped drive to nasbox.
Carried out a restore.
Booted server, bingo, all working. I now have extra unallocated space.
Extended 2nd partition (data), now has loads of space.


Things went well, only problem I had was, it wasn't my test server (ML110 G5) like the customers, it was a Fujitsu Primagy 100. Good thing is I do have another job coming up where I need to so it on same hardware.
Another thing is it only left me with unallocated space, I now need to see if I can extent the partitions. I can just move data and remap network drives but it I needed more room for the OS it would have left me with another task.
 
1) Install additional 2x1TB drives
2) Create new RAID1 array with them
3) Ghost old array to new
4) Remove old disks and boot from the new

Btw. Sin_Chase, it's possible to out of that scenario by using a Backup Exec Intelligent Disaster Recovery CD. You can use the disk manager within there to reassign the drive to C:
 
You must log in or register to reply here.
Back
Top Bottom