Colin Brace wrote:
On 7/18/06, Paul Howarth <paul@xxxxxxxxxxxx> wrote:
In what way did your partition get "hosed"?
Carelessness on my part. I was configuring a new drive and I
accidentally ran mkfs on an existing partition rather than the new one
I had created moments before.
It looks like the filesystem on the logical volume got hosed too.
Indeed. I am perplexed though, as the instructions given on that Red
Hat page seem to offer a way of restoring the undamaged partitions.It
says specifically:
"Assuming a physical volume has been lost that was a part of a volume
group (for what ever reason), the following procedure can be followed.
The procedure will add another physical volume to the volume group to
restore the volume group status and access any data that may be able
to be recovered."
I think the key here is "that may be able to be recovered."
If you have a volume group split over three disks, and a bunch of
logical volumes in that volume group, it's possible that many of those
logical volumes will not have any extents mapped to one particular
physical volume, so if you lose that particular physical volume, the
logical volumes that aren't mapped to that physical volume will be
recoverable.
On the other hand, if you just have one big logical volume filling the
volume group, you're going to lose the data from that volume if you lose
any of the drives.
I can restore the contents of the LVM from backups, but it concerns
500 GB, so I'd rather not I if I don't need to. Surely is must be
(should have been?) possible to restore the physical volume(s) that
didn't get damaged. If not, I'll have to seriously consider whether
using LVM is worth the trouble.
What would you use instead? In what way would that survive the loss of a
drive? You could use RAID with redundancy but that would cost you capacity.
Paul.
