vzdump problem - unable to detect lvm volume group

[xkuba]

Hi,

I'm having problem with vzdump. After machine reboot, it can't use lvm snapshots for backups:

px# pveversion --verbose
pve-manager: 1.5-6 (pve-manager/1.5/4655)
running kernel: 2.6.18-2-pve
proxmox-ve-2.6.18: 1.5-5
pve-kernel-2.6.18-2-pve: 2.6.18-5
qemu-server: 1.1-11
pve-firmware: 1.0-3
libpve-storage-perl: 1.0-9
vncterm: 0.9-2
vzctl: 3.0.23-1pve8
vzdump: 1.2-5
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm-2.6.18: 0.9.1-5

px# vzdump --snapshot --storage backup --maxfiles 10 137
INFO: starting new backup job: vzdump --snapshot --storage backup --maxfiles 10 137
INFO: Starting Backup of VM 137 (qemu)
INFO: running
INFO: status = running
INFO: mode failure - unable to detect lvm volume group
INFO: trying 'suspend' mode instead
INFO: backup mode: suspend
...

This was working before. I can't see anything strange about lvm:

px# lvs
LV VG Attr LSize Origin Snap% Move Log Copy% Convert
data pve -wi-ao 361,26G
root pve -wi-ao 96,00G
swap pve -wi-ao 4,00G

px# df /var/lib/vz
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/pve-data 372864840 10108536 362756304 3% /var/lib/vz

I can create lvm snapshot manually without any problem.

I've tried to look at source code of vzdump but I can't speak perl much.

Any pointers to what to check?

Thanks,

Kuba

 

[dietmar]

What is the output of

# cat /proc/mounts

 

[xkuba]

We've found the root cause. We've changed default locale from C to cs_CZ and that means that decimal point is "," instead of ".". And that breaks vzdump parsing...

 

[blami]

Attached patch fixes problem.

 

[dietmar]

We've found the root cause. We've changed default locale from C to cs_CZ and that means that decimal point is "," instead of ".". And that breaks vzdump parsing...

What environment variable do you set exactly? In vzdump, I set "LANG=C" at startup - I thought this is good enough?