There are several methods to move running Linux to another hard drive at the same server. But I used Unix dump/restore utility to perform this…
First of all it’s necessary to partition new hard drive in the same way as it’s done with old drive (Linux is running at). I usually use ‘fdisk’ utility. Let’s assume that old drive is /dev/hda and new one is /dev/hdb. To view hda’s partition table please run ‘fdisk -l /dev/hda’ which should show something like this:
Disk /dev/hda: 60.0 GB, 60022480896 bytes
255 heads, 63 sectors/track, 7297 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/hda1 * 1 15 120456 83 Linux
/dev/hda2 16 276 2096482+ 82 Linux swap
/dev/hda3 277 7297 56396182+ 83 Linux
After this run ‘fdisk /dev/hdb’ and make the same partitions at it. Interactive mode of fdisk utility is well documented and is very intuitive, so I don’t think it would be difficult to perform partitioning.
After this is done, we should make new filesystems at partitions we’ve created:
mkfs -t ext3 /dev/hdb1
mkfs -t ext3 /dev/hdb3
mkswap /dev/hdb2
When it’s done it’s NECESSARY to mark newly created filesystems as it’s done with old ones. To check filesystem volume name run command ‘tune2fs -l /dev/hda1 | grep volume’ and etc. You’ll see something like this:
Filesystem volume name: /boot
It means that we should mark new hdb1 with label /boot. It can be done by command:
tune2fs -L “/boot” /dev/hdb1
The same should be performed for all partitions except swap one. In my case I should label hdb3 by command:
tune2fs -L “/” /dev/hdb3
At this point new hard drive preparation is finished and we can proceed with moving Linux to it. Mount new filesystem and change directory to it:
mount /dev/hdb1 /mnt/hdb1
cd /mnt/hdb1
When it’s done we can perform moving by command:
dump -0uan -f – /boot | restore -r -f –
And the same with / partition:
mount /dev/hdb3 /mnt/hdb3
cd /mnt/hdb3
dump -0uan -f – / | restore -r -f –
When dump/restore procedures are done we should install boot loader to new HDD. Run ‘grub’ utility and execute in it’s console:
root (hd1, 0)
setup (hd1)
quit
In case everything is done carefully and right (I’ve tested this method by myself) you can boot from new hard drive and have ‘old’ Linux running at new hard drive running.
Good luck!
Hi, good info, but using dump & restore is very proprietary to each filesystem. It will work for Linux's Ext2/Ext3 filesystems, but it may not work if you're using another Linux filesystem like ReiserFS or JFS or XFS. The dump program is proprietary to each filesystem, though the restore is portable. An alternative solution would be to use tar in a similar respect to how you used the dump and restore:
tar cf – –one-file-system /source | ( cd /destination ; tar xvf – )
The first tar command in the sequence acts like your dump, while the second tar inside the parenthesis "()" act like your restore. The first tar sends its output directly to the second tar through the standard input and output. No tar archives are created.
The "–one-file-system" option only works with GNU tar (which is the version that comes with Linux), and it prevents the tar from crossing filesystem mount points. This way you will only backup and restore one file system at a time.
Hi, nice guide, but i'm stuck at the "dump" part, do you think you could help me out?
oaeide@server:/mnt/sdd1$ dump -0uan -f – /boot | restore -r -f –
DUMP: You can't update the dumpdates file when dumping a subdirectory
DUMP: The ENTIRE dump is aborted.
restore: Tape read error on first record
Both Sda1 and Sdd1 are labaled /boot
this is very useful to us.
Thanking you,
with regards,
p.sankar
I realize this post is very old, but i was wondering if i could move my debian to another hard drive, but keep booting grub from the old one. I need this because i’m trying to move debian to an ide hard disk and my server only boots the scsi drives. Is it possible to load grub in one disk and have the actual os on another?
The last sentence in the post is pretty useful 🙂