LVM1 -> LVM2 with failed disk

I've been running debian lvm10 on my machine because 2.4.22 didn't have dm support and I couldn't figure out how to get the patches working.

I've since upgraded to 2.4.27 but never upgraded to lvm2. I do run in on two other machines.

Of course, one of my disks died. In the past, I used the -P partial flag to get the lvm to start so I could copy off my data from the "good" pv, but under lvm10, I don't see that option.

What are my options. Can I upgrade from lvm10 to lvm2 with a failed disk. The lvm doesn't start and while I can pvdisplay /dev/hda4 I cannot vgdisplay or lvdisplay (obviously) since the vg isn't starting.

Can I put in a dummy drive, somehow tell lvm to pvcreate a pv on the new drive with the same ID and then get the vg to come up?

Any other ideas. I'd like to salvage the data that is physically on the working drive.

Thanks

Hi All!

This latter option worked for me once. The VG's name is "vg0", the died PV was /dev/hdc1.

* Read the exact size of the died PV:

vgcfgrestore -ll -n vg0

* Add a new disk to the system, create a partition with _exactly_ the same size as the died PV. (It need not be on the same device as the old PV, eg. I'll call it /dev/hdb2.)

* Restore the LVM headers of the died PV onto the new disk:

vgcfgrestore -n vg0 -o /dev/hdb2 /dev/hdc1

* Rescan, activate the VG:

vgscan; vgchange -a y vg0

Hope it will help.

norbi