Upgrade Disks in a RAID

[IKWeb]

Hi All

 

I have a server which has 6 disks in, all disks are 250GB, 2 of these disks hold the OS, (Mirrored) and the other 5 are setup as RAID 5.

 

Space is now running out and I am looking to buy bigger disks, for the DATA side only, so my question is, if I remove one of the 250GB disks from the RAID 5 and replace this with say a 2TB disk, and then do this with all the others would this migrate the data over?

 

Or would I have to backup all the data, remove all 4 250GB disks, install new disks, and then setup a new VD in the RAID card, and then copy my data back again??

 

Any new would be very welcome.

 

TIA

[plexer]

If you replace the drives one at a time it will have to rebuild the array each time, when you've replaced them all you would be able to expand into the new space.

 

However each rebuild is a risky process in itself.

 

Backup everything including raid config then install new drives.

 

Ben

[jamesfed]

I'd go for a complete backup and then restore to a newly created RAID array - rebuilding the RAID5 array onto new disks each time would take agggesss and you would have a significant performance hit while it goes on.

[plexer]

If you can backup the existing raid config from the card do that also so that your backout plan is - put all original disks back in and restore raid config.

 

Ben

[Arthur]

If I remove one of the 250GB disks from the RAID 5 and replace this with say a 2TB disk, and then do this with all the others would this migrate the data over?

This process is quite risky with large drives and RAID-5. See this article for a good explanation. As you can see from the figures below, SATA drives are much more likely to have read errors during a rebuild compared to enterprise SAS drives.

 

[b]Hitachi DeskStar 7K3000[/b] (2TB)   1 sector in every 11.3686838 terabytes (10^14 bits) will have a non-recoverable read error
[b]Samsung SpinPoint F4 EG[/b] (2TB)   1 sector in every 113.686838 terabytes (10^15 bits) will have a non-recoverable read error
[b]Seagate Cheetah 15K.7[/b] (600GB)   1 sector in every 1.11022302 petabytes (10^16 bits) will have a non-recoverable read error

 

I have a server which has 6 disks in, all disks are 250GB, 2 of these disks hold the OS, (Mirrored) and the other 5 are setup as RAID 5.

2 + 5 = ?