nottelling

no. Arp bridges layer 1 and 2. It's switch local. With a VLAN, it becomes VLAN local, in the sense that 802.1q creates a "virtual" switch.

Broadcast traffic (such as DHCP) doesn't cross subnets without a router configured to forward it. It's one of the reasons subnets exist.

What in the world is "a proprietary OS I cannot trust". What's your actual threat model? Have you actually run any risk analyses or code audits against these OSes vs. (i assume) Linux to know for sure that you can trust any give FOSS OS? You do realize there's still an OS on your dumb switch, right?

This is a silly reason to not learn to manage your networking hardware.

A VLAN is (theoretically) equivalent to a physically separated layer 2 domain. The only way for machines to communicate between vlans is via a gateway interface.

If you don't trust the operating system, then you don't trust that it won't change it's IP/subnet to just hop onto the other network. Or even send packets with the other network's header and spoof packets onto the other subnets.

It's trivially easy to malform broadcast traffic and hop subnets, or to use various arp table attacks to trick the switching device. If you need to segregate traffic, you need a VLAN.

Edit: Should probably note that simply VLAN tagging from the endpoints on a trunk port isn't any better than subnetting, since an untrusted machine can just tag packets however it wants. You need to use an 802.1q aware switch and gateway to use VLANs effectively.

What you are asking will work. That's the whole point of subnets. No you don't need a VLAN to segregate traffic. It can be helpful for things like broadcast control.

However, you used the word "trust" which means that this is a security concern. If you are subnetting because of trust, then yes you absolutely do need to use VLANs.

The answer to your overarching question is not "common maintenance procedures", but "change management processes"

When things change, things can break. Immutable OSes and declarative configuration notwithstanding.

OS and Configuration drift only actually matter if you've got a documented baseline. That's what your declaratives can solve. However they don't help when you're tinkering in a home server and drifting your declaratives.

I’m pretty certain every service I want to run has a docker image already, so does it matter?

This right here is the attitude that's going to undermine everything you're asking. There's nothing about containers that is inherently "safer" than running native OS packages or even building your own. Containerization is about scalability and repeatability, not availability or reliability. It's still up to you to monitor changelogs and determine exactly what is going to break when you pull the latest docker image. That's no different than a native package.

Just cause you've never seen them doesn't make it not true.

Try using quadlet and a .container file on current Debian stable. It doesn't work. Architecture changed, quadlet is now recommended.

Try setting device permissions in the container after updating to Debian testing. Also doesn't work the same way. Architecture changed.

Redhat hasn't ruined it yet, but Ansible should provide a pretty good idea of the potential trajectory.