can't restore KVM from one host to another - tar: Unexpected EOF in archive

[yaboc]

Hi,

I have two Proxmox hosts

box1 pve-manager: 2.1-1 (pve-manager/2.1/f9b0f63a)

box2 pve-manager: 2.2-24 (pve-manager/2.2/7f9cfa4c)

and i can't move the KVM (25GB in size) over to the newer box (box2).

command vzdump 101 --compress lzo --dumpdir /back completes successfully on box 1 and then i can SCP both files
vzdump-qemu-101-2012_11_12-17_03_26.log vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo to box 2

but when i perform qmrestore vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo 101 on box 2

im getting

extracting archive '/home/vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo'
extracting 'qemu-server.conf' from archive
extracting 'vm-disk-ide0.raw' from archive
Formatting '/var/lib/vz/images/101/vm-101-disk-1.raw', fmt=raw size=32768
new volume ID is 'local:101/vm-101-disk-1.raw'
restore data to '/var/lib/vz/images/101/vm-101-disk-1.raw' (26843545600 bytes)
lzop: <stdin>: Compressed data violation
tar: Unexpected EOF in archive
tar: Error is not recoverable: exiting now
lzop: <stdin>: Compressed data violation
tar: Unexpected EOF in archive
tar: Unexpected EOF in archive
tar: Error is not recoverable: exiting now
868219392 bytes copied, 15 s, 55.13 MiB/s
starting cleanup
temporary volume 'local:101/vm-101-disk-1.raw' sucessfuly removed
command 'zcat -f|tar xf /home/vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo '--to-command=/usr/lib/qemu-server/qmextract'' failed: exit code 2

I know that there's one thread that mentions the cause for this could be that the / partition which restore uses as workspace to extract the KVM is lower than the size of KVM, but in this case KVM is 25GB and my df -h on box2 is

Filesystem Size Used Avail Use% Mounted on
/dev/mapper/pve-root 95G 2.5G 88G 3% /
/dev/mapper/pve-data 1.7T 254G 1.5T 15% /var/lib/vz

Any ideas what could be happening over there ? Thanks in advance.

y

 

[dietmar]

Can you restore on box 1 (or decompress with lzop)?

 

[yaboc]

good point. nope it didn't work. getting the same message when trying to restore with different vmid

and also when doing lzop -d vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo
im getting
lzop: vzdump-qemu-101-2012_11_12-17_03_26.tar.lzo: Compressed data violation

so it the archive corrupted ?

why would command vzdump 101 --compress lzo --dumpdir /back

i forgot to mention that the VM is offline too when i did the backup.

do i need to try different compression format

 

[yaboc]

hhhh

 

[dietmar]

do i need to try different compression format

Try again with lzop to see it the error is reproducible.

What kind of backup storage do you use (is it reliable, or maybe damaged?)

 

[yaboc]

update
the backup worked without setting compression flag, so it did it that way. worked well. thanks

 

[yaboc]

hmm im getting kernel panic - not syncing: Fatal exception when i do vzdump 101 --dumpdir /back on stopped KVM

with the newest kernel 2.6.32.16-pve also with 2.6.32.11-pve

 

[yaboc]

Try again with lzop to see it the error is reproducible.

What kind of backup storage do you use (is it reliable, or maybe damaged?)

everything is local for now.

getting kernel panic if i don't set compression flag on kernels 2.6.32.16-pve, 2.6.32.11-pve and 2.6.32.7-pve

when i use lzop as compression the archive is corrupt when i try to decompress it, same thing with tar.gz

im going to test if hard drive is bad, but it's really strange that it worked once before and now it stopped

 

[yaboc]

ok i tested the hard drive WD with manufacturers utility and it passed both short and long tests. so it's probably not the hardware causing the kernel panic.
if I do regular tar and gzip backup im getting kernel panic with not much info in logs. if i do lzo backup completes but the archive is giving me checksum error when i try to decompress. any ideas ? thanks

 

[dietmar]

Already tested the RAM?