[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[linux-lvm] Erroneously formatted PV; 'vgscan -- ERROR "vg_read_with_pv_and_lv(): current PV" can't get data of volume group'...


two days ago i installed a new HDD and added it to my VG 'vg1'
(partition id '8e' + 'pvcreate /dev/hdb1' + 'vgextend vh1 /dev/hdb1').
A bit later I erroneously formatted '/dev/hdb1' by 'mkfs.xfs /dev/hdb1'.
I ran 'vgscan' and 'pvscan' and the made a smiliar output like it is
shown on
http://lists.sistina.com/pipermail/linux-lvm/2002-August/011947.html -
but i have only two PVs in my VG 'vg1': '/dev/hdc1' and '/dev/hdd2'.
I don't remember what i exactly did with '/dev/hdb' after that (it was
late...), but as a result I 'fdisked' the partition table away...
There was no data on '/dev'hdb' (except the LVM metadata), so no 'real'
data needs to be recovered (at least i hope so).

All in all, I can't access the VG 'vg1' (PVs: '/dev/hdc1' + '/dev/hdd2')
with the 3 LVs 'root', 'storage' and 'swap' any more. 

Now you could advise me to run 'vgcfgrestore', but i regretfully must
say that i rebooted my box and as the name of the LV 'root' implies, my
root-filesystem was on it, so the system doesn't boot anymore *doh*...

- 1 VG: 'vg1'
- 2 (3) PVs: '/dev/hdc1' + '/dev/hdd2' + for a short time (see above)
- 3 LVs: 'root', 'storage' and 'swap'
- Filesystem on all LVs: XFS

Fortunately i have a separated 'backup system' on another HDD, which
doesn't use LVM nor XFS, but simply ext2, so i can work with it
temporarily to fix this disaster.

I think the problem is, that the VG 'vg1' still thinks, it would consist
of 3 PVs, but only 2 are there.
But because no VG is found by 'vgscan' (see link above), i can't remove
'/dev/hdb1' by 'vgreduce vg1 /dev/hdb1' and so on...

Is there any way to generate metadata for '/dev/hdb1' to make it known
by 'vg1' or remove '/dev/hdb1' from 'vg1'?!

I'm very despaired and would be gladdened by every answer.
If you need any details, just ask.

P.S.: This is not intended to be a cross-posting; I only had problems
submitting my first one.

Kind regards and thanks in advance

Tim-Patrick Märk

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]