vzdump snapshot mode can't create lvm only on guests running on second drive

[stra4d]

I have a system running version 1.9 that has two drives. The main drive w/ the OS has two guests and they backup fine (storage for dumps is setup on second drive). However, the VMs running on the second drive cannot dump a snapshot, and get a mode failure - unable to detect lvm volume group, and so don't backup with a snapshot but instead suspend the guest.

vzdump:

vzdump --quiet --snapshot --compress --storage vzdump --maxfiles 1 --mailto scott@rci.ca --all
output of guest that runs on secondary drive:

103: Mar 08 23:35:50 INFO: Starting Backup of VM 103 (qemu) 103: Mar 08 23:35:50 INFO: running 103: Mar 08 23:35:50 INFO: status = running 103: Mar 08 23:35:50 INFO: mode failure - unable to detect lvm volume group 103: Mar 08 23:35:50 INFO: trying 'suspend' mode instead 103: Mar 08 23:35:50 INFO: backup mode: suspend 103: Mar 08 23:35:50 INFO: ionice priority: 7 103: Mar 08 23:35:51 INFO: suspend vm 103: Mar 08 23:35:51 INFO: creating archive '/mnt/drive2/vzdumpbackups/vzdump-qemu-103-2012_03_08-23_35_50.tgz' 103: Mar 08 23:35:51 INFO: adding '/mnt/drive2/vzdumpbackups/vzdump-qemu-103-2012_03_08-23_35_50.tmp/qemu-server.conf' to archive ('qemu-server.conf') 103: Mar 08 23:35:51 INFO: adding '/mnt/drive2/vm/images/103/zmail.qcow2' to archive ('vm-disk-ide0.qcow2') 103: Mar 09 00:21:54 INFO: Total bytes written: 27051826176 (9.34 MiB/s) 103: Mar 09 00:21:54 INFO: archive file size: 16.65GB 103: Mar 09 00:21:55 INFO: resume vm 103: Mar 09 00:21:55 INFO: vm is online again after 2764 seconds 103: Mar 09 00:21:56 INFO: Finished Backup of VM 103 (00:46:06)

~# pveversion --v
pve-manager: 1.9-24 (pve-manager/1.9/6542)
running kernel: 2.6.32-6-pve
proxmox-ve-2.6.32: 1.9-47
pve-kernel-2.6.32-6-pve: 2.6.32-47
qemu-server: 1.1-32
pve-firmware: 1.0-14
libpve-storage-perl: 1.0-19
vncterm: 0.9-2
vzctl: 3.0.29-2pve1
vzdump: 1.2-16
vzprocps: 2.0.11-2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.15.0-1
ksm-control-daemon: 1.0-6

Any ideas?

 

[dietmar]

Please can you post the VM config and the storage.cfg?

 

[stra4d]

/etc/qemu-server/103.conf:

name: zmail
ide2: cdrom,media=cdrom
vlan0: rtl8139=66:C4:06:0F:0F:C1
ostype: l26
memory: 3072
onboot: 1
sockets: 1
cores: 4
boot: dac
freeze: 0
cpuunits: 1000
vga: std
acpi: 1
kvm: 1
ide0: drive2:103/zmail.qcow2
bootdisk: ide0

/etc/pve/storage.cfg:

dir: local
        path /var/lib/vz
        content images,iso,vztmpl,rootdir

dir: drive2
        path /mnt/drive2/vm
        content images

dir: vzdump
        path /mnt/drive2/vzdumpbackups
        content backup

dir: drive2_ISOs
        path /mnt/drive2/iso
        content iso

 

[dietmar]

Maybe storage 'drive2' and 'storage 'vzdump' are on the same logical volume. Or is /mnt/drive2 a NFS share?

 

[stra4d]

/mnt/drive2 is a mounted ext3 filesystem. Can the vm storage and the dump storage not be on the same filesystem?

 

[dietmar]

/mnt/drive2 is a mounted ext3 filesystem.

Ext3 does not support snapshot itself. Snaphots only works if the underlying device is an LVM volume.

 

[stra4d]

Not sure I follow. The snapshots have to be created/stored on a drive that is an LV? If so, why do the two guests who are running on the main hdd not have issues when storing to the same location?

 

[dietmar]

Not sure I follow. The snapshots have to be created/stored on a drive that is an LV?

The underlying block device needs to be a LV, yes.

If so, why do the two guests who are running on the main hdd not have issues when storing to the same location?

Because that is on a LV (/dev/pve/data is a LVM logical volume).

 

[stra4d]

But the snapshots are NOT on a LV. They are just on an ext3 disk. All snapshots are stored to the same place.

Do the guests have to run on an LV?

 

[dietmar]

But the snapshots are NOT on a LV. They are just on an ext3 disk. All snapshots are stored to the same place.

The backup target does not matter in that case.

Do the guests have to run on an LV?

The VM images can reside on a filesystem if the underlying block device is a LV, or the images can be a LV itselfs

 

[stra4d]

The VM images can reside on a filesystem if the underlying block device is a LV, or the images can be a LV itselfs

So, just to be clear, my second/additional hdd must be a logical volume with an ext3 (or whatever fs), not just an ext3 filesystem. Correct?

 

[dietmar]

So, just to be clear, my second/additional hdd must be a logical volume with an ext3 (or whatever fs), not just an ext3 filesystem. Correct?

yes