This module is responsible for creating a k0s cluster control plane HAProxy load balancer configured according to the k0s guidelines using a QEMU virtual machine. This module need only be used when creating a multi-node control plane. It is created using the outputs of the control-plane module in addition to the standard container configuration.
This module's outputs are compatible with the outputs of the control-plane module so that either can be used as the control_plane input of the primary module.
This allows both small, single node control planes as well as larger, multi-node control planes to be used interchangeably.
While "normal" DHCP would not be desireable since IPs could change, it's entirely possible to configre DHCP for static assignments.
Unfortunately, neither of the Proxmox providers support retrieving the IP address from a VM or container that uses DHCP.
The IP addresses are needed by k0sctl for cluster configuration.
Because of this, we must assign IPs statically during VM creation.
The way this has been designed in these modules is that a subnet CIDR block is allocated for each "pool" of VMs, be they controllers or workers.
The CIDR block need not be the entire network, of course.
For example if the network is 10.1.2.0/24, a subnet of 10.1.2.192/29 could be used for a pool of VMs.
Care must taken that any DHCP service on the network will not allocate IPs from the subnet range.
A base index into that CIDR block is then used to select an IP for each VM.
This combination lets us share a single subnet for multiple pools.
For example, if the subnet is 10.1.2.128/26, the controller load balancer could use index 0, giving it IP 10.1.2.128, the controllers (pool of 3) could use a base index of 1, giving them IPs 10.1.2.129-10.1.2.131, and finally a worker pool (pool of 10) could use a base index of 5, giving them IPs 10.1.2.133-10.1.2.142.
Presumably, if you're using this module, you have multiple control plane nodes. If everything is running on a single Proxmox node, then that's all well and good. If you also have multiple Proxmox nodes and are distributing the control plane nodes across Proxmox nodes, you should really configure Proxmox HA for this VM. With the control plane nodes distributed across multiple Proxmox nodes, the HAProxy load balancer becomes the single point of failure.