vzdump / qmrestore error

[nsc]

Hi,

I get error with some kvm when restoring some backup :

/mnt/md2/nfs/backup/vzdump-qemu-104-2010_12_18-20_09_13.tar
INFO: reading archive member 'qemu-server.conf'
INFO: CONFIG: name: XXXXXXXXXXX
INFO: CONFIG: ide2: local:iso/en_win_srv_2003_r2_standard_cd1.iso,media=cdrom
INFO: CONFIG: bootdisk: virtio0
INFO: CONFIG: ostype: w2k3
INFO: CONFIG: memory: 1280
INFO: CONFIG: sockets: 2
INFO: CONFIG: vlan2: virtio=7E:E8:89:75D:5A
INFO: CONFIG: virtio0: local:104/disk01-pve.raw
INFO: CONFIG: boot: cad
INFO: CONFIG: freeze: 0
INFO: CONFIG: cpuunits: 1000
INFO: CONFIG: acpi: 1
INFO: CONFIG: kvm: 1
INFO: CONFIG: onboot: 0
INFO: CONFIG: description:
INFO: CONFIG: cores: 1
INFO: CONFIG: virtio2: sata:vm-104-disk-1
INFO: CONFIG: lock: backup
INFO: CONFIG: #vzdump#map:virtio0:vm-disk-virtio0.raw:19280297984:local:
INFO: CONFIG: #vzdump#map:virtio2:vm-disk-virtio2.raw:85899345920:sata:
INFO: reading archive member 'vm-disk-virtio0.raw'
INFO: 0+17657100 records in
INFO: 0+17657100 records out
INFO: 19280297984 bytes (19 GB) copied, 211.318 s, 91.2 MB/s
INFO: tar: Archive contains `\001\024\001pA\026\001\211A\032\001\360' where numeric off_t value expected
INFO: tar: Archive contains `Ad\001\373\001\254\001\377A\255\001\n' where numeric size_t value expected
INFO: tar: vm-disk-virtio2.raw: invalid sparse archive member
INFO: reading archive member 'vm-disk-virtio2.raw'
INFO: 0+0 records in
INFO: 0+0 records out
INFO: 0 bytes (0 B) copied, 285.994 s, 0.0 kB/s
INFO: tar: Skipping to next header
INFO: tar: Error exit delayed from previous errors
INFO: reading '/mnt/md2/nfs/backup/vzdump-qemu-104-2010_12_18-20_09_13.tar' failed - command 'tar xf '/mnt/md2/nfs/backup/vzdump-qemu-104-2010_12_18-20_09_13.tar' '--to-command=/usr/sbin/qmrestore --info --storage local /mnt/md2/nfs/backup/vzdump-qemu-104-2010_12_18-20_09_13.tar --extract 103'' failed with exit code 2

The facts is my VM use 2 differents storage :
- one "local" based on ext3 (SSD hard disk)
- second is a lvm volume (Sata hard disk)

if i put all the disks on the same storage, i have no errors...

Or it's because my second disk is big (80Gb) ?

A bug may be ?

Nsc

 

[dietmar]

What is the output of:

# pveversion -v

 

[nick]

try to restore the VM to another no on virtual machine:

qmrestore name_of_archive.tar 999

where 999 is the new VM number.

 

[nsc]

vm backup from :

pve-manager: 1.7-10 (pve-manager/1.7/5323)
running kernel: 2.6.32-4-pve
proxmox-ve-2.6.32: 1.7-28
pve-kernel-2.6.32-3-pve: 2.6.32-18
pve-kernel-2.6.32-4-pve: 2.6.32-28
pve-kernel-2.6.24-11-pve: 2.6.24-23
qemu-server: 1.1-25
pve-firmware: 1.0-9
libpve-storage-perl: 1.0-16
vncterm: 0.9-2
vzctl: 3.0.24-1pve4
vzdump: 1.2-9
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1dso1
pve-qemu-kvm: 0.13.0-2
ksm-control-daemon: 1.0-4

restored to

pve-manager: 1.7-10 (pve-manager/1.7/5323)
running kernel: 2.6.32-4-pve
proxmox-ve-2.6.32: 1.7-29
pve-kernel-2.6.32-3-pve: 2.6.32-18
pve-kernel-2.6.32-4-pve: 2.6.32-29
qemu-server: 1.1-27
pve-firmware: 1.0-9
libpve-storage-perl: 1.0-16
vncterm: 0.9-2
vzctl: 3.0.24-1pve4
vzdump: 1.2-10
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1dso1
pve-qemu-kvm: 0.13.0-3
ksm-control-daemon: 1.0-4

103 is "non existant" VM

i tried to restore on another Proxmox server (test one).

for info i remove my 80Gb sata disk, i create a 32Gb stored with the 18Gb and the backup run perfectly... (Backup then qmrestore --info)

 

[dietmar]

for info i remove my 80Gb sata disk, i create a 32Gb stored with the 18Gb and the backup run perfectly... (Backup then qmrestore --info)

Is there a reliable way tp reproduce that bug (exact steps)?

 

[nsc]

creating a vm on server with mixed storage ?

i have a /vz stored on SSD (ext3) and a sata with LVM.

i know, this setup is a bit weird, but i can't go in full lvm at this time ...

 

[dietmar]

creating a vm on server with mixed storage ?

no, that does not reproduce the bug here.

Maybe the bug is related to that specific disk "virtio2:vm-disk-virtio2.raw"?

Does it happen when you create a new (empty) disk?