How to move VM (including snapshots) to a new storage?

[turnicus]

Hello,

I am using proxmox 6.2-4 on a single node. I did not use ZFS encryption at first, but I would like encrypt everything now. Unfortunately, it's impossible to encrypt existing ZFS datasets, so I just created new encrypted datasets to which I want to move my VM's now. So my storage configuration looks like:

ID	Type	Content	Path/Target	Shared	Enabled	Bandwidth Limit
hddpool-type_zfs	ZFS	Disk image, Container		No	Yes
hddpool-type_directory	Directory	VZDump backup file, ISO image, Snippets, Container template	/hddpool/proxmox_directories/type_directory	No	Yes
hddpool-enc_type_zfs	ZFS	Disk image, Container		No	Yes
hddpool-enc_type_directory	Directory	VZDump backup file, ISO image, Snippets, Container template	/hddpool/enc_proxmox_directories/type_directory	No	Yes

• the ID "hddpool-type_zfs" points to the non-encrypted dataset hddpool/proxmox_directories/type_zfs and "thin provision" is enabled
• the ID "hddpool-enc_type_zfs" points to the encrypted dataset hddpool/enc_proxmox_directories/type_zfs and "thin provision" is enabled

So I would just like to move my VM's from hddpool-type_zfs to hddpool-enc_type_zfs. I tried using the "Move disk" button (VM -> Hardware -> Select HDD -> Move Disk) but it doesn't move the snapshots.

Could you please explain me how I can move EVERYTHING to the encrypted datasets?

Thanks for any help!

 

[Dunuin]

Do you really need the snapshots? I always just delete them and then move the the VM discs from one to another pool. If anything doesn't work and I need to restore a previous state I just use my backups.

 

[spirit]

I think the only way to to do it manually with "zfs send|zfs receiv" commands

 

[turnicus]

Hello and thanks for your answers

Do you really need the snapshots?

Yes if possible, hence the whole point of this question!

I think the only way to to do it manually with "zfs send|zfs receiv" commands

This was my first idea and I tested it before creating this thread in this way:

- I moved the disk of the VM 110 with the GUI

- I "zfs send hddpool/proxmox_directories/type_zfs/vm-110-state-snap1 | zfs recv hddpool/enc_proxmox_directories/type_zfs/vm-110-state-snap1"

- I "zfs destroy hddpool/proxmox_directories/type_zfs/vm-110-state-snap1"

- I updated the new path of the snapshot in /etc/pve/qemu-server/110.conf

- I tried to rollback to snap1 with the GUI but I got an error "can't rollback, more recent snapshots exist".

I guess I should do it exclusively through the command line but what am I missing?

 

[ness1602]

I would also really love to move VM with snapshots but i dont know if it is possible with QEMU/LXC atm.

 

[turnicus]

Hello,

OK, I think I found the correct way.

- step1 = shutdown VM via the GUI

- step2 = create a snapshot (called "ready_to_move" in this example) via the GUI

- step3 = send this snapshot recursively (with the -R flag) which will send the complete tree (snapshot + all previous snapshots + initial volume)

zfs send -R hddpool/proxmox_directories/type_zfs/vm-110-disk-0@ready_to_move | zfs recv hddpool/enc_proxmox_directories/type_zfs/vm-110-disk-0

- step4 = send the state volumes (states volumes are created when snapshotting an active VM with the "include RAM" checkbox ticked). You can list them with "zfs list"

zfs send hddpool/proxmox_directories/type_zfs/vm-110-state-snap2 | zfs recv hddpool/enc_proxmox_directories/type_zfs/vm-110-state-snap2

- step5 = update all paths in /etc/pve/qemu-server/110.conf

- step6 = start VM and test everything

- step7 = destroy the old volumes and snapshots (with the -r flag)

zfs destroy -r ssdpool/proxmox_directories/type_zfs/vm-110-disk-0

- step8 = destroy the state volumes

zfs destroy ssdpool/proxmox_directories/type_zfs/vm-110-state-snap2

Hope it helps!