Is there some way to self host what cloudflare does?

Your domain will always have to be rented through a 3rd party. Cloudflare is (or was?) one of the better choices for that.

Cloudflare does other things as well, most notably it can acts as a proxy: an inbetween between your server and the users. This inbetween can be useful against DOS attacks, blocking of bots, etc. But for most self hosters that part is not necessary. It’s a toggle in cloudflare’s DNS dashboard: I think you’d want it to say DNS only.

Another thing cloudflare can do is tunneling. It’s useful for when your server is behind a firewall or NAT or double NAT you can’t or don’t want to configure. You’d probably know if you use this, so I assume you don’t?

Is there any meta analysis on these major outages?

They seem to be occuring more and more regularly.

Depending on 3rd parties is a pain in the ass

Awesome selfhosted lists https://github.com/heyform/heyform as a possibility.

if a tool exists which can easily and reliably detect AI generated content then they’d just be using that tool for their training

Generative Adversarial Networks are an example of that idea in action.

One of these projects might be of interest to you:

• https://github.com/Mintplex-Labs/anything-llm

• https://github.com/mudler/LocalAI

Do note that CPU inference is quite a lot slower than GPU or the well known SAAS providers. I currently like the quantized deepseek models as currently the best balanced between quality of replies and inference time when not using GPU.

I’ve worked with bookstack. Found it easy and intuitive. It’s wiki software not specific to family history.

I think it depends on the rate of change, rather than the amount of containers.

At home I do things manually as things change maybe 3 or 4 times a year.

Professionally I usually do setup automated devops because updates and deployments happen almost daily.

At one of my clients, who wants everything on-prem, I use gitlab CI with ansible. It took 3 days to setup, and requires thinkering. But all in all, I like the versitility, consistency and transparency of this approach.

If I’d start over again, I’d use pyinfra instead of ansible, but that’s a minor difference.

Stock raspberry os and syncthing sounds like the easiest way to do this.

I also dislike graphana kabana elastic behemot.

You can use rsyslog to centralize the logs. Then there’s tools like this for anomaly detection on those logs.

A small application I wrote myself, hosted on the free tier of pythonanywhere.com

Uptime monitoring and notifications

I’ve done cron @reboot keep-one-running <mycommand> before (1)

Great news, thanks!

Hopefully the android releases will be able to follow, as I understand the original syncthing authors will no longer be supporting android.

Aah, ISP’s NAT. Yes, in that context, it’s correct that you can’t port forward.

Perhaps you can STUN through, but unlikely to get a good port.

Port forwarding was invented for exactly that

1. Port forwarding
2. Yes, and there’s services that do that for you

Static IP is helpfull but not necessary. Even with NAT and a changeing IP there’s options, such as:

1. dynamic dns.
2. Public reverse proxy or tunnel.
3. Onion routing.