Restoring to CEPH

[nz_monkey]

Hi,

I just got my Proxmox VE cluster up and running with ceph. I just tried migrating one of my VM's from local storage to RBD by taking a backup, then restoring it to a new RBD based VM. This unfortunately does not work, the error I get is:

extracting archive '/var/lib/vz/dump/vzdump-qemu-101-2013_01_17-14_17_52.tar.lzo'extracting 'qemu-server.conf' from archive
extracting 'vm-disk-virtio0.raw' from archive
new volume ID is 'ceph:vm-103-disk-1'
restore data to 'rbd:rbd/vm-103-disk-1:id=admin:auth_supported=cephx\;none:keyring=/etc/pve/priv/ceph/ceph.keyring:mon_host=10.8.1.1\:6789\;10.8.1.2\:6789\;10.8.1.3\:6789' (21474836480 bytes)
unable to open file 'rbd:rbd/vm-103-disk-1:id=admin:auth_supported=cephx\;none:keyring=/etc/pve/priv/ceph/ceph.keyring:mon_host=10.8.1.1\:6789\;10.8.1.2\:6789\;10.8.1.3\:6789' - No such file or directory
tar: vm-disk-virtio0.raw: Cannot write: Broken pipe

When will backup/restore using rbd be supported in Proxmox VE ?


The ability to do "live block migrations" as supported by QEMU 1.3 would also be a very welcome addition

 

[dietmar]

When will backup/restore using rbd be supported in Proxmox VE ?

Planned for the next release.

 

[l.ansaloni]

Planned for the next release.

Hi,
when this feature will be implemented?
I have try with release 3.3 but it don't work.
Thanks in advance,
Lorenzo

 

[dietmar]

when this feature will be implemented?
I have try with release 3.3 but it don't work.

This is implemented, but only with new vma backup format. Old tar files cannot be restored to rbd.

 

[Mitya]

You can do it manually for old archives. For images, just untar archive and do rbd import. Then get conf, rename to some free VMID, place in /etc/pve/qemu-server/ and fix parameters inside.

 

[l.ansaloni]

You can do it manually for old archives. For images, just untar archive and do rbd import. Then get conf, rename to some free VMID, place in /etc/pve/qemu-server/ and fix parameters inside.

Thank You for reply,
I resolve in another way:
- add new HDD SATA to one node of the clustrer;
- add to that node a new storage "directory" using the new HDD;
- restore VM into that storage;
- live move the disk to ceph storage;
- remove the HDD SATA.

Lorenzo