Move current local and local-lvm to ezisting local-zfs pool

[insalata_fresca]

Hi guys,
UI am fairly new to proxmox and I am starting now to experiment.
Roght now i have a 1TB SSD where proxmox has been installed with all its lxc and VM and 2 storages on it (local and local-lvm)
I have also a zfs-pool of 4 2TB nvme drives and a storage upon it called local-zfs ehre all my data resides.
I want to remove the 1TB SSD and move local and local-lvm to the zfs nvme pool without having to reinstall everything.

Is this somehow possible?

 

[Dunuin]

Hi guys,
UI am fairly new to proxmox and I am starting now to experiment.
Roght now i have a 1TB SSD where proxmox has been installed with all its lxc and VM and 2 storages on it (local and local-lvm)
I have also a zfs-pool of 4 2TB nvme drives and a storage upon it called local-zfs ehre all my data resides.
I want to remove the 1TB SSD and move local and local-lvm to the zfs nvme pool without having to reinstall everything.

Is this somehow possible?

I'm sure that is somehow possible. But I'm also quite sure that it would be easier and less work to just install everything from scratch.
You would need to somehow shink your ZFS partitions without corrupting the data on the pool in order to get unallocated space to create two new partitions. Then you would need to partition and format them and some how move that root filesystem and bootloader to these new partitions.
If you don't know how to do that (I do not) you will probably screw up your PVE and won't be able to boot and maybe you also lose all your data.

So I personally would backup the /etc folder (and maybe the config.db), back up all VMs and LXCs, wipe all those ZFS disks and then install PVE from scratch and restore the backups.

 

[insalata_fresca]

the problem is that i have all of my data on the ZFS disks so I didnt want to destroy that.
it is somehow possible to do a new installation on another 2TB SSD (it is on the way now) and then join that SSD to the existing ZFS pool extending it?
My goal is to not touch the existing local-zfs pool which has all of my data

 

[Dunuin]

the problem is that i have all of my data on the ZFS disks so I didnt want to destroy that.

Raid isn't counting as a backup and you should always have backups of your data. So to do things properly you would need some additional storage to backup your ZFS pool anyway. With that it wouldn't be a problem to destroy the pool.

I personally always got another same sized pool on another host so I can use ZFS replication to move all the datasets/zvols between the hosts.

it is somehow possible to do a new installation on another 2TB SSD (it is on the way now) and then join that SSD to the existing ZFS pool extending it?
My goal is to not touch the existing local-zfs pool which has all of my data

That would make things easier but really depends on your pool layout. In case you use a raidz1/2/3 your for example can't add a new single disk without destroying and recreating that pool (another point why you might want a second pool for backups, so you can destroy the pool for pool expansion).
Another problem would be that the bootloader + system would be on a single disk, so no redundancy and data loss and a non-working server once that single disk fails.

In case you got a striped mirror (raid10) that should work by adding two more disks and installing PVE to those two disks first leaving the remaining part of the new disks unpartitioned. You could then partition that unallocated space later and add these partitions to the existing pool as a new mirrored vdev.

 

[insalata_fresca]

so the safes solution is to grab my old HDD (i have 2 8TB ones) and set them as cold backup so taht everything on the server can be destroyed anytime and restored from the spinning HDD?

 

[insalata_fresca]

what is the smartes way to full backup my volumes autoamtically to the new backup pool? I would do a simple mirror there.

 

[Dunuin]

so the safes solution is to grab my old HDD (i have 2 8TB ones) and set them as cold backup so taht everything on the server can be destroyed anytime and restored from the spinning HDD?

Yes. Here ia an example on how to copy data between pools on block level:
https://docs.oracle.com/cd/E18752_01/html/819-5461/gbchx.html