K3CAN

The slightly lower power draw pi5 vs a Tiny will eventually make up for the higher initial cost, but you can save more by turning off lights when you leave a room or skipping a round at the bar.

In my opinion, the wider software compatibility, better processing power, and expansible RAM and storage options far outweigh the eventual theoretical savings.

That said, if you need the super small SBC form factor or GPIO pins, definitely go for a pi. They absolutely have their use cases. I have 4 or 5 of the 3B and 3B+, and have used them on-and-off for a variety of tasks over the years.

It’s been a long time since Pi’s were competitive on price.

You can get a used Lenovo Thinkcentre for $50 on eBay. A modern pi is going to cost you that much for just the board, then you still need to buy a case, power supply, SD card, and then figure out some solution for storage…

No reason to bother with a Pi unless you need the GPIO for something. You can do more with a Lenovo Tiny or SFF Dell.

Host? As in running services?

Wireguard and the Proxmox Backup Server software itself. Redundancy/failover comes from the server cluster itself, not my backup server.

As far as the backup content, it “hosts” backup images of my VMs and LXCs, plus /home from my laptop in case it ever gets lost or damaged.

I tried a bunch, zoneminder, motioneye, frigate, etc., before finally settling in AgentDVR. It offers a fair bit of flexibility via MQTT and “just worked” with my PTZ camera.

You’re not a “target” as much as you are “a thing that exists.” These aren’t targeted attacks.

That said, you can look into adding some additional measures to your webserver if you haven’t already, like dropping connections if a client requests a location they shouldn’t, like trying to access /admin, /…/…, /.env, and so on.

On nginx, it could be something like:

location ^/\.|)/admin|/login {
    return 444;
}

Of course, that should be modified to match whatever application you’re actually using.

Self hosted from my homelab on an nginx server. I also self host my blog, which has some info on my whole set up. My blog uses some basic bloging software, though, rather than being hand-made.

The “side menu thingy” is achieved through HTML “frames”. It’s an element of HTML that’s pretty much extinct nowadays, but was all the rage when I built my very first page back in the day.

Nice. I wrote mine “by hand”, too. No CSS, just raw HTML. I think it’s a more personal experience than just using whatever random template some all-in-one web hosting company offers.

A lot of how you set up your system is just going to depend on how you want to set it up.

I run podman (like an improved version of docker) in a single LXC container for applications that are primarily packaged as docker apps. I think I have 4 or 5 applications running on that LXC.

For things that are distributed via apt, git repo, etc, I’ll either create a new LXC or use an existing LXC if it’s related to other services I’m running. For example, crowdsec is run in the same machine as nginx since those two work together and I’ll always want them both running at the same time, so there’s no reason to separate them.

I have mariadb running in its own LXC so that it can follow a different (more frequent) backup schedule than the mostly static applications that interact with it.

Anything that needs to interact directly with hardware, like Home Assistant, or I want kernel separation for, will get a full fledge VM instead of a container.

It’s all about how you want to use it.

I use podman almost exclusively at this point. I like having the rootless containers and secrets management. If you’re on Debian, though, I strongly suggest pulling podman from Trixie. The version in Bookworm is very out of date and there’s been a lot of fixes since then.

For what it’s worth, though, you can proxy other services, like Gemini or gopher, through the same proxy for simplicity’s sake.

I self host.

I use nginx as a reverse proxy with crowdsec. The backends are nginx and mariadb. Everything is running on Debian VMs or LXCs with apparmor profiles and it’s all isolated to an “untrusted” VLAN.

It’s obviously still “safer” to have someone else host your stuff, like a VPS or Github Pages, etc, but I enjoy selfhosting and I feel like I’ve mitigated most of the risk.

I’d imagine that if your job is making YouTube videos, portainer and other graphical abstraction layers probably make more visually interesting videos than just watching someone type out a bunch of commands.

If you’re going to be playing with custom locations and such, it might be worth using nginx directly instead of through the limitations of NPM.

I know I’m a bit late to the conversation, so I don’t know if this is still helpful… But I have a camera with “AI Detection” built into it and it appears to send alerts via its ONVIF connection. I’ve disabled motion and other detectors on my NVR (AgentNVR) and instead configured it to just wait for an alert from the camera itself to start recording. It’s been working quite well.

My initial plan was to use a coral TPU and frigate, but the Coral/Gasket drivers appear to be pretty old and I couldn’t get them to work properly, myself.

I’ve also been running nginx in an unprivileged LXC container. I haven’t used fail2ban, specifically, but crowdsec has been working without issue.

You can mostly just treat an LXC like a normal VM.

I know you said you decided against it, but perhaps reconsider USB?

I was facing the same dilemma a few months ago, and ultimately decided that trying to break out those internal connections wasn’t worth it. The problem with these tiny PCS is that they are not designed with arrays of drives in mind. There’s typically not enough room in the case to properly add an additional drive, so you end up running the sata cord through a hole in the case and using an external drive and power supply anyway.

USB on the other hand, is intended to connect to an external device. The connectors themselves are more robust and they can even supply power.

I use my external drive for data I don’t have to access constantly, like templates and backups. 90% of the time it’s just sitting in standby. If you need to access a lot of data constantly, you might start to notice the slower USB speeds; if you can segment your data, keeping your “working” files on the internal drive and just use the external for storage, you probably won’t notice the USB at all. It’s certainly not the perfect solution, but with your particular restrictions, it might be the better tool for the job.

The other option would be Network Attached Storage, essentially a low power computer that just exists to allow other computers to access its storage. You can probably find an old, cheap desktop PC for sale locally, likely for less than it would cost you to bring out those internal ports and buy a drive enclosure and power supply.

I’m using cloudflare as my nameserver and the free API seems to work just fine with ddclient.

I do occasional smaller “takeouts” and haven’t had any issues.

I have an “automatic album” (or whatever they call it) where all the photos of friends and family (even pets) get automatically added to it. Then I can just request a “takeout” for that one album, since those are the photos I actually care about. It’s a much smaller download than the entirety of my Photos account.

I have a couple Libre Office files where I document the non-technical stuff for my own quick reference, like network layout in Draw, or IP and port assignments in Calc. I use a git repo to store and organize podman scripts, systemd unit files, configs, etc. Probably not the most elegant solution, but it’s simple and FOSS.

Reverse proxy is Nginx Proxy Manager.