Moving the VM's 1 Node to another with ZFS

[Naresh Avunoori]

A cluster with 3 Nodes with ZFS Storage on Each node.

1. To Utilize the Maximum storage opted for ZFS.

Question: Is there any possibilities to MOVE the VM's from 1 Node to another, may be in Between can be have a NFS.

NOTE: Tried with [VM --> DiskAction --> Move Storage]; able to move the Disk to NFS, and NFS to another Node. but failing to copy the configurations.

 

[itNGO]

There is no need for NFS... You can just live migrate between the nodes if CPUs are compatible or offline migrate. ZFS will do the "replica-magic" then....

 

[Naresh Avunoori]

Thank you,

I don't have same Names for the ZFS Storages in the DataCenter, and not allowing me to keep the same name.
so added, Nodes to the Node1's Storage ID: Node1_Storage --> ZFS Pool: Node1_localPool --> Nodes: Node1, Node2 and also Node2's Storage ID: Node2_Storage --> ZFS Pool: Node2_localPool --> Nodes: Node2, Node1

Below is the Replication log I see:
Header
Proxmox
Virtual Environment 8.2.4
Virtual Machine 100 (testvm) on node 'node1'
No Tags
Logs
()
2025-04-15 11:37:10 100-0: start replication job
2025-04-15 11:37:10 100-0: guest => VM 100, running => 0
2025-04-15 11:37:10 100-0: volumes => Node1_Storage:vm-100-disk-0,Node1_Storage:vm-100-disk-1
2025-04-15 11:37:11 100-0: (remote_prepare_local_job) zfs error: cannot open 'Node1_Storage': no such pool
2025-04-15 11:37:11 100-0: (remote_prepare_local_job)
2025-04-15 11:37:11 100-0: (remote_prepare_local_job) zfs error: cannot open 'Node1_Storage': no such pool
2025-04-15 11:37:11 100-0: (remote_prepare_local_job)
2025-04-15 11:37:11 100-0: (remote_prepare_local_job) could not activate storage 'Node1_Storage', zfs error: cannot import 'Node1_Storage': no such pool available
2025-04-15 11:37:11 100-0: (remote_prepare_local_job)
2025-04-15 11:37:11 100-0: end replication job with error: command '/usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=Node2' -o 'UserKnownHostsFile=/etc/pve/nodes/Node2/ssh_known_hosts' -o 'GlobalKnownHostsFile=none' root@192.168.1.x -- pvesr prepare-local-job 100-0 Node1_Storage:vm-100-disk-0 Node1_Storage:vm-100-disk-1 --last_sync 0' failed: exit code 1

 

[itNGO]

Post in thread 'Renaming ZFS Storages in a Cluster for Replication'

Mar 25, 2025

Pretty much that with one small hint:

Updating each VM configuration to reflect the new ZFS pool name

The VM config doesn't necessarily reference the pool name, but the name of the storage. But the storage name is often the same name as the pool.

Make backups of the guests before you attempt this. And of course, the guests need to be powered down while you do this. Or alternatively, migrate the guests to the other node. Then you could skip the manual change in the VM configs.

• aaron

• 

 

[UdoB]

Is there any possibilities to MOVE the VM's from 1 Node to another,

I don't have same Names for the ZFS Storages

When I select a running "VM --> context menu --> Migrate" I get a dialog popping up. On the right side I have two drop down menus: "Target node:" and "Target storage:" where I can select another... storage as the destination on the receiving node.

In my setup a stopped VM does not offer this choice.

A cluster with 3 Nodes with ZFS Storage on Each node.

1. To Utilize the Maximum storage opted for ZFS.

Oh! Interesting. For the integrated compression?

I am a ZFS fan and I use it wherever I can. (With acceptable technical effort and an eye on the available money.)

But "maximize storage" was never my main point. With three nodes and storage replication and at least a mirror for each vdev of the pools... every block of a VM gets stored six times on six similar devices. Plus the space used on two primary and three secondary PBS', also on redundantly configured ZFS


Again: I use ZFS everywhere I can, just for completely different reasons.