Why move fastmail? They are Australian.

For Dropbox, I replaced it with syncthing completely, works great

I was wondering mainly about TLDs because for registrat I think even OVH would be ok

Many companies use different domain for China, literally .com.cn.

Damn! I was literally wondering this on my run today, and also... how stable/independent of US are which TLDs? Can they take over easily .com? (Not block, but steal)

My email is on .de, probably safe? There's .eu... Unsure who's running that.

I need this, but on my proxmox cluster or PC, with network firewall via... Instant snapshot and restore should be easy via ZFS/BTRFs.

someone***

I'll figure out at some point how to add new udev rule based on nobuto-m.github.io/post/2025/ho... but really? That's not accessible to any normal person.

Everything is better with Niri too! Some settings are per output only and it works seamlessly on my 40" and 13" screen.

I only wish I could sync some login session automatically and not have to log in everywhere again

Ok, after reading more about frequency of Fedora updates, I decided to stick with Ubuntu. Claude iterated on the script a little bit, added more volumes, etc., and now I'm running.

When setting up new machine I write "manual" notes and this year there aren't that many to set up new system!

Yep, just with tests in between and fully automated rollbacks. Claude is handling nix pretty well. There's a lot of good quality open source nix code!

Another idea that is still back on my mind.

Use 2-3 "cheap" servers for DB and web, connect from home server only for workers and async Tasks. After all, we're running on BEAM, right? Basically, hosting would cover only cases where latency to the user matters, the rest offloaded.

That's not a lot of resources and probably would require lots power too? Plus single point of failure...

If I had reliable & fast internet (🇩🇪LOL) I'd rent only cheap hetzner VPS for forwarding to my server at home.

Mac Minis for hosting are interesting too, very power efficient

Unsure if this counts, I have multiple proxmox LXC/VMs deployed via deploy-rs with NixOS. In 2025 I mostly switched to just prompting Claude with what I need and letting it run tests, checks and if all ok - deploy.

It was important to get healtchecks right, and deploy-rs handles auto rollbacks.