I see, do you know of a way in Docker (or Podman) to bind to a specific network interface on the host? (So that a container could use a macvlan adapter on the host)

Or are you more advocating for putting the Docker/Podman containers inside of a VM/LXC that has the macvlan adapter (or fancy incus bridge adapter) attached?

Confused at this sentiment, Docker includes a MACVLAN driver so clearly it’s intended to be used. Do you eschew any networking in Docker beyond the default bridge for some reason?

With the default Docker bridge networking the container won’t have a unique IP/MAC address on the local network, as far as I am aware. Communication with external clients will have to contact the host server’s IP at the port the container is tied to in order to interact. If there’s a way to specify a specific parent interface, let me know!

This was very insightful and I’d like to say I groked 90% of it meaningfully!

For an Incus container with its unique MAC interface, yes if I run a Docker container in that Incus container and leave the Docker container in its default bridge mode then I get the desired feature set (with the power of onions).

And thanks for explaining CNI, I’ve seen it referenced but didn’t fully get how it’s involved. I see that podman uses it to make a MACVLAN interface that can do DHCP (until 5.0, but the replacement seems to be feature-compatible for MACVLAN), so podman will sidestep the pain point of having to assign a no-go-zone on the DHCP server for a Docker swath of IPv4s, as you mentioned. Close enough for containers that the host doesn’t need to talk to.

So in summary:

• I’ve got Docker doing the extent it can manage with MACVLAN and there’s no extra magicks to be done on it.

• Podman will still use MACVLAN (no host to container comms still) but it’s able to use DHCP to get an address for the MACVLAN container.

• If the host must talk to the container with MACVLAN, I can either use the MACVLAN bypass as you linked to above or put the Docker/Podman container inside an Incus container with its bridge mode.

• Kubernutes continues to sound very powerful and flexible but is definitely beyond my reach yet. (Womp womp)

Thanks again for taking the time to type and explain all of that!

Thanks for taking the time to reply!

The host setup has eth0 as the physical interface to the rest of the network, with br0 replacing it completely. br0 has the same MAC as the eth0 interface and eth0 just forwards to br0 which then does the bridging internally. br0 being a bridge means that incus is able to split it off without MACVLAN but rather its nic device in bridge mode which “Uses an existing bridge on the host (br0) and creates a virtual device pair to connect the host bridge to the instance.” That results in a network interface that has its own MAC and is assigned a local IP by the DHCP server on the network while also being able to talk to the host.

Incus accomplishes the same goal as Proxmox (Proxmox has similar bridge network devices for its containers/VMs) just without Incus needing to be your OS/distro like Proxmox does, it’s just a package.

As for the Docker, the parent interface is br0 which has supplanted eth0. MACVLAN is working as it is intended to in Docker, as far as I can tell. The container has a networking device with its own MAC address, and after supplying the MACVLAN network device with my network’s subnet and gateway and static IP address in the Docker compose file it works as expected. If I don’t supply a static IP in the Docker compose file, Docker just assigns it the first IP in the given subnet - no DHCP interaction. This docker-net-dhcp plugin (I linked to the issue about it not working on the latest version of Docker anymore) was made to give Docker network devices the ability to use DHCP to get an IP address, but it’s clearly not something to rely on.

If I’m missing something about MACVLAN that makes DHCP work for Docker, let me know! Hardcoding an IP into a docker-compose file adds an extra step to remember compared to everything else being configured on the centralized DHCP server - hence the shoddy implementation claim for Docker.

Thanks for the link to using another MACVLAN and routing around the host<-/->container connection issue inherent to MACVLAN. I’ll keep it in mind as an alternate to Incus container around another container! I do wish there could be something like Incus’ hassle-free solution for Docker or Podman.

Not what you asked for but possibly useful; if you have apple devices and can use airplay instead of Bluetooth, https://github.com/mikebrady/shairport-sync works really well. Even runs airplay 2 on a pi zero smoothly. Don’t know of Bluetooth otherwise sadly

Sad to hear for my quadlet future, do you remember what things were specifically annoying?

Hey bigdickdonkey, I recently tried and wasn’t able to shit my way through podman, there just wasn’t enough chatter and guides about it. I plan to revisit it when Debian 13 comes out, which will include podman quadlets. I also tried to get podman quadlets to work on Ubuntu 24 and got closer, but still didn’t manage and Ubuntu is squicky.

I read about true user rootless Docker and decided that was too finicky to keep up to date. It needs some annoying stuff to update, from what I could tell. I was planning on many users having their own containers, and that would have gotten annoying to manage. Maybe a single user would be an OK burden.

The podman people make a good argument for running podman as root and using userns to divvy out UIDs to achieve rootless https://www.redhat.com/en/blog/rootless-podman-user-namespace-modes but since podman is on the back burner till there’s more community and Debian 13, I applied that idea to Docker.

So I went with root Docker with the goals of:

• read only
• set user to different UID:GID for each container
• silo containers in individual Docker networks
• nothing gets /var/run/docker.sock
• cap_drop: all
• security-opt=no-new-privileges
• volumes all get tagged with :rw,noexec,nosuid,nodev,Z

Basically it’s the security best practices from this list https://cheatsheetseries.owasp.org/cheatsheets/Docker_Security_Cheat_Sheet.html

This still has risk of the Docker daemon being hacked from the container itself somehow, which podman eliminates, but it’s as close to the podman ideal I can get within my knowledge now.

Most things will run as rootless+read-only+cap_drop with minor messing. Automatic ripping machine would not, but that project is a wild ride of required permissions. Everything else has succumbed, but I’ve needed to sometimes have a “pre launch container” to do permission changes or make somewhere like /opt writable.

I would transition one app stack at a time to the best security practices, and it’s easier since you don’t need to change container managers. Hope this helps!

Your budget is really near a https://store.ui.com/us/en/collections/unifi-dream-router/products/udr Unifi dream router. Your family is gonna be way happier with you (0 downtime) and it’ll give you extender options if you ever need it. Unifi is good enough and they update regularly, just disable cloud access stuff and you’re good.

Otherwise you want Opnsense instead of Openwrt. The upgrade process for Openwrt is not automatic, while Opnsense is. Worth it not to have to dote on your router.

And you should get an access point (Unifi something or Tplink Omsomething), wifi is problematic with openwrt and I’m not sure if opensense even lets you do it (haven’t tried).

And you’ll need a switch, dumb or managed, up to you if you want VLANs. The Opnsense box will have just one LAN port, so it requires a switch if you want to plug more than one thing into it. A switch with PoE+ can power the access point directly.

Opnsense needs x64 arch (Intel or AMD CPUs), get a small thin client like a Dell Wyse 5070 extended or HP T730 or that mentioned Fujitsu Futro S720 (its CPU is old tho, you can do better). There may be newer thinclients, you just want a mini PCIe slot to install some Intel gigabit card from eBay with 2 ports. Google power efficient gigabit mini PCIe card - there’s an older model that sucks power and a newer one that doesn’t suck; if you go more than gigabit skip 2.5 on Intel unless you google hard and expect extra power draw. Very limited point to 4 port cards, just go higher gigabit speeds don’t think about multiplexing ports or whatever it is called; and switches switch better than the router can and remove CPU overhead for more actual routing work - 2 port card is the way.

Slap Incus (superior but newer, less guides, LXD is previous name if googling stuff) or Proxmox (good enough, more guides for this) on it, make a VM and pass through the 2 ports of the PCIe cards, slap Opnsense in the VM. Make an LXC container and slap Debian on it and spin up the Unifi controller for your AP. Another container for adguard home or pi hole and you’ve got a box that does the basic nets all in one. The built-in port on the thin client is how you will access the underlying OS, it gets plugged into the switch you’ll have to get. If you got something with 2 gigs of RAM and an AMD Geode/GX or aged Intel Atom CPU I’d just only do Opnsense no hypervisor stuff.

Sorry for the info dump but there’s a lot of angles!

But really, the Unifi dream router is much easier and solves it all-in-one. You need 3 pieces (router, wifi access point, Ethernet switch) for a good experience otherwise.