It’s not true.

But if it was, it wouldn’t be a reason to dislike Postgres. Academic AI research in the '90s is a far cry from commercial AI companies today.

Thanks!

What’s the reference here?

Insomnia > Postman.

I switched to Insomnia around 2021 when Postman started enshittifying and found I liked it a lot more. Insomnia has also been relatively enshittified unfortunately, but it feels like it’s to a lesser extent.

Your 142.x.x.x will be your public IP address. All devices on your network share that public IP. They all have a unique private IP address too, accessible only on your network. It probably starts with 192.168.x.x, but it could be 10.x.x.x or even less likely 172.16–31.x.x.

If you want to operate a web server that users can go to by typing https://youdomain.com/, you’ll need to forward from ports 80 and 443 through to the internal IP address of your server, using the “port forwarding” settings on your router. What port on the internal IP you route to depends on how your server is configured. But a basic default configuration is fairly likely to be 80 and 443, too.

Since you have a reverse proxy, all traffic from your router should go to that. Then you use that to send the appropriate traffic to the appropriate server based on whatever rules you want to apply. (e.g. siteone.mydomain.com goes to server 1, sitetwo.mydomain.com goes to server 2, or mydomain.com/siteone goes to server 1, etc.).

Oh good tip!

That’s a pretty righteous set up OP.

Lol not me. I’m not the author. Just saw the article and thought it was an interesting conversation starter.

or a recipe for an insecure mess that could become difficult to maintain

The concept, or the specific setup the author of that article has? If you mean the latter, I’m not going to argue. But the concept? It shouldn’t have any effect either way on security, but the whole advantage of it is that it’s less of a mess. The same way that running a whole bunch of services on bare metal can quickly become a mess compared to VMs or Docker/LX containers, declared state helps give a single source of truth for what all the services you might be running are. It lets you make changes in repeatable and clearly documented ways, so you can never be left wondering “how did I do that before?” if you need to do it again.

If everything you run is a Docker container, there’s a good chance Terraform is overkill; a Kubernetes config will probably do the job. But depending on your setup there are a whole bunch of different tools that might be useful.

What’s your preferred approach to defined state in your home servers?

Oh, I used HA to mean high availability. I was not aware people also abbreviated Home Assistant.

Sorry for the late reply. I’m just disorganised and have way too many unread notifications.

LXC containers sound really interesting, especially on a machine that’s hosting a lot of services. But how available are they? One advantage of Docker is its ubiquity, with a lot of useful tools already built as Docker images. Does LXC have a similarly broad supply of images? Or else is it easy to create one yourself?

Re VM vs LXC, have I got this right? You generally use VMs only for things that are intermittently spun up, rather than services you keep running all the time, with a couple of exceptions like HomeAssistant? What’s the reason they’re an exception?

Possibly related: your examples are all that VMs get access to the discrete GPU, containers use the integrated GPU. Is there a particular reason for that distribution?

I’m really curious about the cluster thing too. How simple is that? Is it something where you could start out just using an old spare laptop, then later add a dedicated server and have it transparently expand the power of your server? Or is the advantage just around HA? Or something else?

Sorry for the late reply. I’m just disorganised and have way too many unread notifications.

LXC containers sound really interesting, especially on a machine that’s hosting a lot of services. But how available are they? One advantage of Docker is its ubiquity, with a lot of useful tools already built as Docker images. Does LXC have a similarly broad supply of images? Or another easy way to run things?

Oh yeah, the “run headless” tip too was great! I would never have used a desktop environment, and would in effect have been using it headless. But had you and others not specifically suggested running it as headless it would probably not have occurred to me that that’s a setting change I’d need to make while installing it.

Thanks! I genuinely wasn’t sure how much RAM would be necessary, and would have probably seriously considered 8 GB enough if I hadn’t gotten the feedback.

I’ve no comments on RISC-V, but I agree that a move towards ARM in the Windows & Linux worlds would seem sensible. I would guess it hasn’t happened for the same reason IPv6 hasn’t taken over. Too much momentum. Too many developers still working in an x86 world, too many legacy apps that won’t easily run on ARM, too many hardware manufacturers each making the individual choice to keep making the current-popular option. Apple could transition because they’re the single gatekeeper. They make the decision, and everybody else who wants to use a Mac has to follow along. I’m going to guess that the control they have over the hardware and the software also means Rosetta 2 works a hell of a lot better than Microsoft’s Prism. (I can’t say for sure, having never used an ARM-based Windows machine or an ARM-based Mac.)

In terms of heat, what kind of room do you have it in? Somewhere with good natural airflow, or away in a closet somewhere?

I guess I have the same question for you as I did for curbstickle. What’s the advantage of doing things that way with VMs, vs running Docker containers? How does it end up working?

Interesting. I’ve never really played around with that style of VM-based server architecture before. I’ve always either used Docker (& Kubernetes) or ran things on bare metal.

If you’re willing to talk a bit more about how it works, advantages of it, etc., I’d love to hear. But I sincerely don’t want to put any pressure and won’t be at all offended if you don’t have the time or effort.

I’m not sure I agree with your definition of walled garden. I’d say it’s a place that’s designed to be nice and easy to use within the bounds designed for you (the garden), but which protects the user from doing something that might harm them, even if that “protection” comes at the cost of being able to do other things they want to do, in a kind of paternalistic way (the wall). The classic example would be iOS, where the only apps you can install are the ones Apple has approved for you. Getting apps from the open web the way you would on Windows, macOS, or Linux could be dangerous!

Your description of:

you may run into roadblocks doing things that way, yes. You are pretty much limited to what’s on their (vast) catalog

Makes it sound very much a walled garden to me. Not as high-walled as iOS of course, but it’s a spectrum.

But anyway, it’s basically semantics. Not that important what you call it.

Not at all. It’s completely open source

Being open source doesn’t necessarily preclude being a walled garden. If (and I fully admit I could be completely wrong about this) it makes it easy to do certain things through a friendly UI, but it becomes much harder or more awkward (or impossible) to do things that aren’t explicitly supported, as part of a deliberate design decision/tradeoff for that usability.

Anyway, thanks a heap for answering all my questions. Has been very helpful.