[linux-lvm] vgcfgrestore fails.

[Heinz J. Mauelshagen]

On Sat, Apr 28, 2001 at 12:33:58PM -0500, root wrote:
> I am running SuSe 7.1 
> 2.4.0 kernel 
> 512M Ram
> AhA2940U2W
> suse lvm
> 
> I had three volume groups and a week ago the volume group that contained user 
> had a drive crash.  I was unable to save usr but now I can not mount any of 
> my other volume groups and vgcfgrestore fails.  Is their any thing I can do 
> to get my data?  
> thor:~ # pvdisplay /dev/sdd2
> --- Physical volume ---
> PV Name               /dev/sdd2
> VG Name               export
> PV Size               8.3 GB / NOT usable 3.31 MB [LVM: 238 KB]
> PV#                   2
> PV Status             available
> Allocatable           yes (but full)
> Cur LV                1
> PE Size (KByte)       4096
> Total PE              2123
> Free PE               0
> Allocated PE          2123
> PV UUID               ruLId0-1JoY-34Ja-oLW0-FmzV-Q5ZN-NQaZw9
> thor:~ # pvdisplay /dev/sdc2
> --- Physical volume ---
> PV Name               /dev/sdc2
> VG Name               export
> PV Size               8.3 GB / NOT usable 3.31 MB [LVM: 238 KB]
> PV#                   1
> PV Status             available
> Allocatable           yes (but full)
> Cur LV                1
> PE Size (KByte)       4096
> Total PE              2123
> Free PE               0
> Allocated PE          2123
> PV UUID               ynnKCX-mjlT-jeNe-9MHY-tnP2-XCIC-nbKrmq
>  
> thor:~ # vgcfgrestore -n export /dev/sdd2
> vgcfgrestore -- ERROR: different structure size stored in "/etc/lvmconf/export.conf" than expected in file vg_cfgrestore.c [line 
> 122]
> vgcfgrestore -- ERROR "vg_cfgrestore(): read" restoring volume group "export"
>                                                                               

You are using a newer LVM versions which has different metadata definitions
that the one which created those backups.

My guess is, that you are using LVM > 0.9.1 Beta 3 *now* but created the
backups with a lower LVM 0.9 version and you suffer from a PV uuid related
bug preventing vgscan to find your VG.


One way to address the situation (pressuming no VGs are active) is:

- create /etc/lvmtab.d/ in case it doesn't exist

- copy /etc/lvmconf/export.conf to /etc/lvmtab.d/export

- echo -ne 'export\0' > /etc/lvmtab

- vgchange -ay

If the VG named "export" comes back to life this way (assuming that the user LV
belongs to it) do:

- lvreduce -l1 /dev/export/user
- lvrextend -l1 /dev/export/user

replace "user" with a valid LV name in case I assumed wrong.

The purpose of that NULL operation in the end is, that your PV uuid list
gets recreated by LVM >= 0.9.1 Beta 4 which should make vgscan happy again ;-)

The last 2 commands just shrink user by 1 LE and grow it again which
is necessary, because your export VG is *full*. This will not harm the user
LV in the end, because the allocator has the only option to use the very same
PE for growing which was freed before by the shrinking of user.

If your VG had at least 1 free PE a lvcreate/lvremove cycle for a dummy
LV with just 1 LE had done it as well.

Please tell us if this helps you or we need to go for another solution.

>           
> _______________________________________________
> linux-lvm mailing list
> linux-lvm at sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

-- 

Regards,
Heinz    -- The LVM Guy --

*** 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
Mauelshagen at Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-