Skip to contentSkip to conversation

pvmove killed vg

linux-lvm [email protected] 2 participants 2 messages Dec 11, 2001 – Dec 12, 2001

lore
Share & tools
lore
#0 lore raw

Hi,

I've got a broken lvm vg as a result of an unsuccessful pvmove command. My system is a SuSE 7.0 with kernel 2.4.4 and lvm-0.9.1_beta7-10.

The vg spans several partitions on one disks, one of which I wanted to evacuate. I used the pvmove command with chunks of 32 PE (= 1 GB); 10 GB total. The first 8 chunks worked okay, but the 9th (pvmove -v /dev/hda8:256-287 /dev/hda10) fell over with error code 23 while moving the fifth PE. So I checked using pvdisplay that the four PE had indeed been moved. A further attempt with pvmove -v /dev/hda8:260-287 /dev/hda10 fell over as well, again error code 23. (That's error moving physical extent(s)).

I then rebooted the machine which killed the complete volume group. Which in turn did not quite impress me. :-((

During my attempts at recovering I think I did a vgscan, which found the vg "vg1" on /dev/hdb, but not the vg "vg0" on /dev/hda.

I tried the following commands unsuccessfully:
- vgcfgrestore -v -n vg0: please enter physical volume name (the synopsys of vgcfgrestore did not state it as a mandatory
parameter)
- vgcfgrestore -v -n vg0 /dev/hda10: can't restore part of active volume grout vg0
- vgcfgrestore -v -n vg0 /dev/hda{6,7,8,10}: please enter physical volume name
- vgchange -a n vg0: volume group vg0 does not exist
- vgchange -a y vg0: volume group vg0 does not exist

I still have the file /etc/lvmconf/vg0.conf.cd as well as the devices /dev/vg0/lv0{0,1,2,3} and /dev/vg0/group. (And /dev/vg1/..., but that is working.)

Any suggestions on how I can recover the data would be highly appreciated. I found a reference to a program called uuid_fixer/uuid_editor, but usage was discouraged. Do I have to try it
or is there a better way? Does upgrading to a newer version of lvm help? Or do I have to update the kernel as well then? And I definately
do _not_ want to kill my other vg as well.

Regards,

Gunther

______________________________________________________________________________
Die schönsten Ski-Regionen der Alpen - jetzt bei Ferienklick.de
http://ferienklick.de/ski/?PP=2-5-100-105-38

#1 lore raw

Gunther,

you need to run vgcfgrestore on *all* PVs which where in your vg0 *after* running "pvcreate -ff" on them. YOu can find out which these where by "vgcfgrestore -ll -n vg0 -f /etc/lvmconf/WhateverYourRecentBackupFileIs".

If that doesn't work a hack to activate it anyway without running vgscan is to copy /etc/lvmconf/WhateverYourRecentBackupFileIs (I assume /etc/lvmconf/vg0.conf.cd in your case) to /etc/lvmtab.d/vg0, "echo -en 'vg0\0vg1\0' > /etc/lvmtab" and "vgchange -ay". Create a dummy LV with 1 PE afterwards and your
metadata on all PVs of the VG should be ok again.

Remember to back /etc/lvmconf/ regularly when you change your LVM configuration!

Regards,
Heinz -- The LVM Guy --

We recommend to upgrade to LVM 1.0.1 because a couple of bugs (some related to pvmove) have been fixed.

Quoted 58 lines · On Tue, Dec 11, 2001 at 10:34:24PM +0100, [email protected] wrote:

Hi,

I've got a broken lvm vg as a result of an unsuccessful pvmove command. My system is a SuSE 7.0 with kernel 2.4.4 and lvm-0.9.1_beta7-10.

The vg spans several partitions on one disks, one of which I wanted to evacuate. I used the pvmove command with chunks of 32 PE (= 1 GB); 10 GB total. The first 8 chunks worked okay, but the 9th (pvmove -v /dev/hda8:256-287 /dev/hda10) fell over with error code 23 while moving the fifth PE. So I checked using pvdisplay that the four PE had indeed been moved. A further attempt with pvmove -v /dev/hda8:260-287 /dev/hda10 fell over as well, again error code 23. (That's error moving physical extent(s)).

I then rebooted the machine which killed the complete volume group. Which in turn did not quite impress me. :-((

During my attempts at recovering I think I did a vgscan, which found the vg "vg1" on /dev/hdb, but not the vg "vg0" on /dev/hda.

I tried the following commands unsuccessfully:
- vgcfgrestore -v -n vg0: please enter physical volume name (the synopsys of vgcfgrestore did not state it as a mandatory
parameter)
- vgcfgrestore -v -n vg0 /dev/hda10: can't restore part of active volume grout vg0
- vgcfgrestore -v -n vg0 /dev/hda{6,7,8,10}: please enter physical volume name
- vgchange -a n vg0: volume group vg0 does not exist
- vgchange -a y vg0: volume group vg0 does not exist

I still have the file /etc/lvmconf/vg0.conf.cd as well as the devices /dev/vg0/lv0{0,1,2,3} and /dev/vg0/group. (And /dev/vg1/..., but that is working.)

Any suggestions on how I can recover the data would be highly appreciated. I found a reference to a program called uuid_fixer/uuid_editor, but usage was discouraged. Do I have to try it
or is there a better way? Does upgrading to a newer version of lvm help? Or do I have to update the kernel as well then? And I definately
do _not_ want to kill my other vg as well.

Regards,

Gunther

______________________________________________________________________________
Die schönsten Ski-Regionen der Alpen - jetzt bei Ferienklick.de
http://ferienklick.de/ski/?PP=2-5-100-105-38

Quoted 5 lines

_______________________________________________
linux-lvm mailing list
[email protected]
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen Sistina Software Inc.
Senior Consultant/Developer Am Sonnenhang 11
56242 Marienrachdorf Germany
[email protected] +49 2626 141200

                                                       FAX 924446

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-