New options are available in the proxmox provider plugin so:
- Add vmid as defined in proxmox today (so it won't try and generate one for some reason)
- Open multiple configuration variables needed: bridge, balloon, numa, ...
- Avoid default values fallback. So some properties are not set for all nodes (e.g. hypervisor, bridge, etc...). They must be declared explicitely for each node instead.
- Had to inline the "module node" call [1] for the storage0 node (2 disks to declare). It's now inlined like the gateway (declares 2 networks). So it's a bit repetitive but it's declared appropriately.
- Separate staging nodes (from production) to its own arborescence. So we don't touch the production nodes when applying
- As a poor man configuration shared management, added the staging terraform.tfstate within the staging arborescence... (no sensible information in there plus now we can share that state)
- Ignore the target_node/hypervisor change (as required in the diff)
Now the` terraform plan` for staging looks reasonable.
I modified some production nodes according to their current configuration but
stopped as it's a subject for another time. (Shout out if i need to remove
those but this sounds like lost time to do so, might as well keep it since it's
done)
[1] The module node only allows 1 disk and 1 network declaration. At the time
of its design, determining how to do conditional/loop were not found.