hello, if your datastores use SAN storage them yes. So ideally in this simple setup mount SAN storage to /var/lib/one and backup your /etc/one dir. You can also use Ansible
are you running Opennebula Front-end or node, or both, on the first machine ?
I take that you run one front (on bare metal) and node (for VMs) on it, as assumption a).
“All the data on SAN” means one installation data + VM data files ? Or just VM data files ?
I take just VM data files as assumption b).
With assumptions a) and b), if you lose machine 1, the VM images will stay on the SAN, and if you re-build your one front and node on new hardware, you will face the following situation:
after recovery of your one front DB, and restart, the VMs will be there
most VMs should be able to restart, but will ave lost their active status: it’s like a reboot after crash
However, beware of one thing: all VMs with non-persistent images will lose their changes EVEN WHEN SAVED ON HARD DISK in case of a re-deploy event.
So you have to make sure that when you recover/rebuild your one front and node, you keep all the same front and node identification (name and ip addresses and ssh keys) so that you do not need to re-deploy VMs to a node seen as a new one.
So beware of always deploying VMs as persistent (or based on persistent image ) if you plan to save important data to remain in case of loss of the server.
Note: if you inadvertently created a VM as non-persistent and still want to save its data or make it persistent, there is also a procedure to do so:
power-off the VM (DO NOT undeploy it)
us the save VM button to create a new VM template with a copy of the image
make sure the newly copied VM image is set as persistent
You now have a copy of your VM template and image that should be persistent.
Just a few hints from lessons I learnt the hard way.