The router port to the proxmox machine is set up for tagged packets that isolate incoming/outgoing traffic.
After that my VMs and Containers are easily set up to "live" on one or more networks.
For me the firewall rules on the router determine what traffic can be relayed between vlans through the router.
I'm pretty sure you could set up opnsense running in a container or vm to do the same thing, selectively passing traffic from one vlan to another.
I agree with OMV. It certainly can be used as is, but not usually how people want to use it. A note was added to the script a few days ago.
> I don't see any glue scripts to get VMs talking to each other
There is a Tailscale script which technically helps them talk to each other (over Tailscale) :)
The scripts are designed to setup self contained LCX containers. We are trying to avoid building our own k8s.
I think I will stick to using proxmox virtual ports to create my network so I can more easily only stick to individual device registration in tailscale and save on that overhead when I'm home, but then also add tailscale /headscale into the mix somewhere so I can tap in via VPN when I am out of the house.
Tailscale and OPNsense are more difficult to get working together due to conflicting project goals (one blocks well, the other opens up well), but it looks like it's worth it to me.
I'm confused by what you mean here? Don't they just use the network like any other computer?
I haven't had to do any special configuration to get my VMs to talk to each other.
And of course this means that the Proxmox box as a whole should have similar hardening to a typical web server, with minor tweaks to allow residential traffic on various other standard ports. So that hardening would probably be another script I would like to see (I don't know what all the proxmox scripts in the first section do).
Since selecting the bridge for a service's NIC is part of setting up each service, the only thing such a "glue script" would be doing is creating the `vmbr1` bridge. That's already a one-liner.
In the router case, you'd likely want this default one to be the 'internal' network and have a separate interface (either physical or VLAN) for the WAN.
OK, so you want to virtualise a router and firewall. That's fine. I have deployed roughly 200 pfSense firewall/routers as VMs and physical boxes and OPNSense is similar, so I can probably help.
At a minimum you will need two physical interfaces (one will actually do but you will need to know what you are doing!). You need "WAN" and "LAN". OPNSense is still FreeBSD based, I think, so it will not run in a L[inux]XC container for obvious reasons.
Your last paragraph seems rather confused. I don't know what you mean by "glue scripts". VMs communicate via networks
I suggest you try a few experiments to get to grips with virtualisation properly and then move on from there. If you swing by the Proxmox forums with specific issues we'll try to help out but in the end you need to dive in full on ... or not.
I wanted to do what I think is a very basic and very common setup: Modem > proxmox box > OPNsense VM > physical wifi router via onboard 10Gb NIC + internal network VMs like OMV etc. The goal is to add a full network filter via OPNsense, and allow access to a media sever and backup etc from the internal network.
I see no OPNsense, OMV script is basically contra-indicated because it should be a VM instead of the LXC container, and I don't see any glue scripts to get VMs talking to each other, which is an important part of Proxmox configuration. So it looks like there is room here to get some basic setup scripts for a simple home server either improved or added to the collection.