manwichmakesameal

That's an understatement. $300 for the TS3+. Holy shit.

This is also dependent on who is running the server and how it's set up. If it's the matrix.org you're using, I couldn't tell you. If it's someone hosting/you're self-hosting, you need a STUN server for traversing NAT. It's not part of the default Synapse docker install and I'm not sure about non-docker installs.

Also, the docker-container/compose version needs sometimes some inter-docker intervention. If you’re not tech-savy and have no idea of python scripts and how to exec into containers, this can sometimes create some frustration, when all engines are raising errors !

I 100% agree that SearxNG is amazing. It's the only engine I use now. I'm not sure I understand how you needed the knowledge of Python though. I've been running SearxNG for quite a while and have never had to touch a single line of Python.

Having your ISP do your port forwarding seems alien to me as that's not the norm where I am. Since it seems like a standard thing where you are, you may run the risk of another ISP doing the same thing. Personally, if the price is right, I'd take the latency hit and get a VPS and route all inbound traffic through that via wireguard.

Yep, this is what I do with my "media acquisition" stuff. I have Jackett, Sonarr, Radarr, Transmission all run from a single compose file.

Are you me? I'm currently running 2 x R710s with an SA120 stuffed with drives. One R710 handles the storage and provided NFS storage for my other R710 running Proxmox with all my VMs/containers. I've been seriously considering downgrading my hardware to some lower power used former SFF workstations.

To add more, I think you're right in suspecting your ./well-known setting. I'm not sure where you would set that in a direct to Matrix setup like you have but it's pretty easy using nginx. I just ran mine through the federation checker and my hosts section came back as exactly what I set in my location /.well-known/matrix/server directive in my nginx config.

So, my setup is this: Nginx reverse proxy using LE wildcard certs for my domain. I put my Matrix on a subdomain on my domain. Then I just point the CF tunnel to the reverse proxy using the subdomain for my public hostname. Everything works for me. Federation, local chat, etc. I have Home Assistant notifying me via Matrix now too.

I'm not sure I understand this. All you have to have for Matrix is a homeserver.yaml file and you can spin up a docker image no problem. Are you saying more expensive resource wise?

How exactly is your Matrix server running? Behind an nginx reverse proxy or are you using the cloudflare tunnel directly to it? Personally, I have mine behind a reverse proxy and just point the tunnel at my proxy.

Same here. I kept using the docker run commands that usually show up in Docker hub but started making my own compose files. So much cleaner feeling. I can keep everything all nice and neat in a single folder now. Makes backing up much easier too.