With the one requirement that you need to pick an DNS server which doesn’t mess with the results. But that’s not a huge issue, there are quite some uncensored DNS servers out there. Like the OpenNIC ones for example.

I know, and some of them have quite some latency, while others may or may not start censoring and/or logging my requests in the future. The downside is that having my own DNS server outside my LAN doesn’t make much sense, because (you’re right!) my home internet connection is struggling to keep up with the major providers, as is my hardware. However, at, I’ve just checked, an average of just under 2,000 requests per hour, it’s more than feasible.

There’s a good chance that the operators won’t even notice my requests, especially because my cache is filling up fast.

And I think it’s really a shame that lots of ISPs mess with the DNS results and introduce third-party blocklists. Mine does that, too.

This. Very much this.

Its robust and fairly straight forward to setup IMO.

I never got it working reliably on OpenBSD, something always resets its root directory’s permissions to root:root which makes the service break. It’s probably unfair of me to blame Unbound for this, but it always sticks in my mind. In addition, Unbound wants a text file as configuration and the solution I have now found does not. It’s also a question of convenience, at least a little. :-)

I’m not ruling out the possibility that I could recreate my local setup with Unbound (there should be a way to automatically download and integrate an AdBlock filter list somehow), but I admit that I’m just not familiar enough with it. It’s a bit of a shame, I know, but unlike a mail or web server, I have really big problems when my DNS server goes down with a cryptic error message. I would like to minimise this risk.

Ah, sorry. This is a new technology to me. Thank you!

I mean that, if all DNS servers just returned whatever the root servers tell them, nobody would want to run his own one, I think.

For the record, any DNS server you choose to employ should default to only using the root servers.

If that was the case, there would be a market for exactly 0 DNS servers.

You should also be aware that even if you use root servers, a DNS server which is authoritative for the domain you are querying may well return different results depending on where in the world you are.

Thank you, this is valuable information to me. :-)

Blocky is another nice AdBlocker and DNS proxy.

Blocky does look nice, but there is no way to use it without any upstream DNS server, which limits its usefulness. Technitium works without having to rely on third-party DNS services, which is its main selling point, I guess.

Yes, it’s running on a Raspberry Pi. Only internal SD, no other services. RPi 3B, I think. It’s been mine for quite a few years now.

Well, VPS then.

What would be a “not personally setup”?

Both, somewhat. It is a virtual root server. I’m still considering to consolidate - at least - my OpenBSD servers into one, but I’m lazy.

My backup server is the only one of my servers that is located outside Germany. You know, in case the British come again. Or the data centre of my other servers burns down. Or something like that.

Every night, this server receives a (compressed, incremental) backup of the most important data (content and configuration files) from each of my other servers, which I created with Borg.

I haven’t been able to try Linkding on any of my servers yet, as neither OpenBSD nor OmniOS are supported. I would probably like it if it did.

1. There is no obvious way to install Hoarder on an unsupported platform. My servers run OpenBSD and OmniOS. Both of them don’t even have any Docker support. (Which is not something I’d absolutely need, to be honest.)
2. Hoarder runs on Node.js. I will use Lisp (edit: or Rust, but I’m positive I’ll beat Lisp’s session management some time soon). Node.js is a dependency hell.
3. Hoarder does not really encourage manual and/or regex-based tagging, it strongly suggests relying on “artificial intelligence”. As I am rather disappointed by what “artificial intelligence” is currently able to do, I’d prefer to default to the old approach.

Of course, all of this is just a personal preference.

Working on my better Hoarder alternative. Goal: maybe 2025.

Honestly, what I use is not what I would recommend. ;-) My own setup follows these directions (TL;DR: OpenBSD, as much OOTB OpenBSD software as possible, and Dovecot.)

I am happy to recommend it to others.

If they ever support non-Docker systems again, I might be curious. Right now, I couldn’t even use that.

Note that you don’t know what the hosters know, store and/or sell about you.

I’m glad that you see my point that “other people hosting your data” is not really a good idea.

They host software for anyone to use, and capture all the data, usage patterns, etc, for themselves, to use for their benefit, and to use against you.

So I guess that we can agree that data stored on other people’s computers will not be safe. I honestly wonder why you think other people’s computers are safer if you know their names.

And you want to sit here and tell me they’re the answer?

I would be very grateful if you would only judge what I have written and not what you think I might have meant.

Are you just an apologist for FAANG, etc?

There is no reason to attack me personally, my friend.

Who’s paying you to post this disinformation?

Just in case I’m fundamentally misunderstanding your personal attack so I don’t report it to the moderators without cause: What is ‘disinformation’ about my pointing out that Google and Facebook host software for other people (even if they have their own motives)?

every community has somebody in it who does the self-hosting for the community

That’s what (e.g.) Google and Facebook do: Host software for the community.