lvm over iscsi vm migration between unclustered nodes

[albertodega]

I have a 3 pve nodes (2 in v3.1, 1 in v3.4).
Every vm on the first 2 nodes are resident on a shared nas, connected to the node via an lvm over iscsi(one per vm) setup on the nodes.
At the moment, alle the nodes are indipendent one form another.
i would like to make a cluster(not in ha) of those 3 nodes(+1 coming next week).

The idea is the following:
1. Migrate all the vm of node 1 to node 2 and 3
2. update node 1 to pve3.4(or 4, as per suggestions)
3. keep moving vms on the 3 nodes until same version reached, then build the cluster between those nodes, having the nas in iscsi as the shared storage.

first question:
which is the better way to move vms from one node to the other with this configuration?
second question:
is the process here stated good to reach a 4 nodes cluster in the end or should i proceed differently?

thank you in advance!
alberto

 

[albertodega]

please, could anyone help here?

 

[Richard]

I have a 3 pve nodes (2 in v3.1, 1 in v3.4).
Every vm on the first 2 nodes are resident on a shared nas, connected to the node via an lvm over iscsi(one per vm) setup on the nodes.
At the moment, alle the nodes are indipendent one form another.
i would like to make a cluster(not in ha) of those 3 nodes(+1 coming next week).

The idea is the following:
1. Migrate all the vm of node 1 to node 2 and 3
2. update node 1 to pve3.4(or 4, as per suggestions)
3. keep moving vms on the 3 nodes until same version reached, then build the cluster between those nodes, having the nas in iscsi as the shared storage.

first question:
which is the better way to move vms from one node to the other with this configuration?

* Precondition: On all nodes the network storage should be defined using the same name (if not more manual adaptions in the following would be necessary)

* Stop the VM(s)

* save the VM(s)'s config data (/etc/pve/qemu-server/<vm-id>.conf ) and delete them (but NOT the virtual disk!) from the current node

* restore the VM(s)'s config data on the target node

* start the VM(s)

* Attention: ensure that there is no contradiction between the config data and the node's resources (network, CPU)

second question:
is the process here stated good to reach a 4 nodes cluster in the end or should i proceed differently?

Yes.

 

[albertodega]

thank you richard!
i've just one remaining question:
if i try to add the lvm over iscsi target on the second node i receive an error stating that the target is already connected to an lvm:

create storage failed: device '/dev/disk/by-id/scsi-360014052c39edd0da77dd3640dba88d0' is already used by volume group 'Impero' (500)

so i'm able to connect the iscsi target, but unable to mount the lvm on it on the second node.
suggestions?

again, thank you in advance

 

[albertodega]

sorry, made it, just an uppercase issue between nodes!