Hello,
I’m looking for guidance on what’s required to modify OpenNebula’s LVM datastore driver so it can operate with local storage instead of shared SAN.
Right now the LVM driver is designed for shared SAN volumes exposed are as documented in SAN Datastore — OpenNebula 6.10.4 documentation .
What I want instead:
-
The Front-End connects to the target hypervisor via SSH
-
Copies the image to the HV over SSH
-
Creates the LV inside a predefined
vg-one-<ds-id>volume group -
Deploys the VM locally on that LV
-
Live or cold migration should still work between HVs (as long as each HV has its own
vg-one-<ds-id>)
My question:
What are the actual technical requirements or components I need to modify?
Is it enough to extend the TM/LVM transfer scripts, or are there additional dependencies in the image manager, drivers, or migration hooks?
I want to understand the full scope before I start patching anything.
Thanks!