Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Due to a necessary upgrade to larger disks in a Dell server I'd like to image the existing CentOS 5.3 installation and recover this onto new disks.

Because the existing disks are RAID-1 and take up both slots in the machine I can't simultaneously have the new and old disks in at the same time. It's a Dell R300.

I can take server down and I'm comfortable using rescue disks for CentOS, and creating partitions etc, just a question of the data transfer. It'll likely either be to a USB disk or over NFS to another server (NFS probably preferred, as Gigabit so faster).

I presume this is something like running rsync/tar with the right options on the various partitions - anyone done this and can suggest suitable commands to achieve it?

thanks

Barnaby

share|improve this question
add comment

5 Answers

up vote 1 down vote accepted

How about CloneZilla? Would that suit you?

Hope this helps, Best regards, Tom.

share|improve this answer
add comment

Why not break the RAID-1 and stick in one of the new disks? If you're willing to take downtime anyway, it's not as though an inopportune disk failure would cause anything awful to happen.

share|improve this answer
    
Thanks for the response. If I understand from reading docs, I'm not sure you can create a new RAID-1 array in a degraded state - the new array would need both new disks in the machine. We've considered replacing one, rebuilding and replacing the other, but this will take 4+ hours, and there's no guarantee of being able to increase the RAID disk size at the end (docs are quiet on this topic!). –  bee Nov 19 '09 at 11:37
    
Worst case (if you're using the PERC RAID), you put both the new disks in, create the RAID, then break that, but Linux software RAID is far more flexible. I don't understand your issue with the disk size; you'd create the new RAID array at it's larger size right from the start. –  womble Nov 19 '09 at 11:53
    
Ah ok - the RAID controller is an LSI based SAS6i/R 'RAID for dummies', if that makes a difference, seems somewhat more primitive than a proper PERC. If I removed both original disks is there a risk the controller could forget the config, ie. not remember both parallel RAID-1 disks? This would be ideal if this could work :-S –  bee Nov 19 '09 at 12:05
    
I don't think even that bodgy thing is dumb enough not to put the RAID metadata on the disks -- because otherwise hardware failure (controller or otherwise) would kill the RAID set permanently. –  womble Nov 19 '09 at 12:58
    
Good point, I'll try this - with suitable backups! Thanks. –  bee Nov 19 '09 at 14:05
add comment

It may be overkill but I used to love systemimager for this kind of work - especially for deploying cloned systems over and over again:

http://wiki.systemimager.org/index.php/Main_Page

share|improve this answer
    
Cheers for the tip, I'll take a look. –  bee Nov 19 '09 at 11:39
add comment

I've done this. With Dell's RAID controller, the RAID volume appears as a single disk to the O/S. This allows most any imaging software to treat it as a single disk. I used G4L (can be used to image any O/S, not just Linux) to image to a nearby FTP server or USB disk, swap out disks, rebuilt RAID volume, and then image back to the upgraded disks.

If you image the entire RAID volume, it will do the partitioning for you. However you will have exactly the same size partitions as before, which may be limiting if you are replacing with larger disks. You can add additional partitions later or bump out the last partition with gparted. With foresight, putting Swap and /var as the last partitions allows some more flexibility here.

The alternative is to image one partition at a time. That is more tedious, but allows you great flexibility in laying out your new partition scheme- as long as each new partition is the same size or larger than the backed-up partition!

share|improve this answer
    
Thanks, G4L looks like just the ticket. –  bee Nov 19 '09 at 14:06
add comment

To update this question with a successful outcome, I ended up using a combination of the above advice. I picked this solution because some downtime was fine in my situation (I was more concerned with spending less time at the data centre!) and wanted to be definitely sure it'd work.

  • Remove both existing drives
  • Add both new drives and create new RAID-1 on these.
  • Remove one of the new drives and replace with one of old - you now have 2 degraded RAID-1 arrays, both accessible.
  • Boot clonezilla CD and clone local disk-disk (g4l didn't look to be able to do direct local-local clone).
  • Remove old drive and replace with remaining new drive. Start rebuild of array. Boot up OS.
  • ..sometime later.. repartition with parted/fdisk/resize2fs

... profit! :-)

cheers for the help!

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.