On Friday, April 14, 2006 6:12 PM +0100 Timothy Murphy
<tim@xxxxxxxxxxxxxxxxxxxxxx> wrote:
cd /mnt/newdisk
dump 0f /dev/sda1 | restore xf -
Thanks.
This sounds plausible - I'll see if it works.
On thinking about it, should it not be
cd /mnt/newdisk
dump 0f - | restore rf -
?
Better be sure before I start ...
Oops, you're almost right. dump needs to know which partition to back up,
and what file to store the backup in. The "f -" says to send the backup to
stdout (ie. the pipe) but you also need to specify the partition to back
up. Recall that dump backs up the raw device for the partition, not the
mount point. That's why it can back up stuff you normally couldn't see due
to other mount points. It can also maintain sparse files sparsely. (I think
tar is now able to do this.) restore, OTOH, restores the backup to a
mounted filesystem, not a raw device. (This asymmetry means that a member
of the disk group can perform a backup, but you need to be root to perform
a verify.)
So I think the correct command is:
cd /mnt/newdisk
dump 0f - /dev/sda1 | restore rf -
And before someone reminds us about Linus saying that dump shouldn't be
used, I'll mention the deprecation FAQ on the dump home page:
http://dump.sf.net/
