sj_zero

It’s based on conduit, which itself is really a lot better for self-hosting than synapse or dendrite.

I ran conduit on an atom d2550 alongside other services and it basically idled. Running synapse or dendrite on the same machine made the whole machine max out permanently.

Irc is fine, but not federated…

My dad found one solution that’s specifically for the task (whose name eludes me at the moment), but for me, my nextcloud is a Swiss army knife.

Apache guacamole is something I wish I had when I started. Let’s you connect with telnet, ssh, RDP, or VNC using html5

In sysctl.conf you can put

kernel.panic=1

Where 1 is the seconds until reboot.

For homeservers if you’re not always on-site to look at it, it’s a good idea to set a reset on kernel panic.

More ideal not to reboot of course, but I am often hundreds of miles from my servers so a kernel panic over something that wouldn’t have otherwise killed the system is something I’d rather live with and reboot.

A Klingon throuple is going to have more black eyes than a bag of peas.

Always go with more ram. I can say that from experience.

I’m partial to fanless, but keep in mind my empire of dirt is almost entirely fanless so I’m just partial to it.

I’m running a 4B model on one of my machines, an old surface book 1.

It’s a brutal machine. heat issues, and the GPU doesn’t work in linux. But pick a minimal enough model and it’s good enough for me to have LLM access in my nextcloud if for some reason I wanted it.

Biggest thing really seems to be memory, most cheaper GPUs don’t have enough to run a big model, and CPUs are dreadfully slow on larger models if you can put enough RAM in one of them.

Hopefully someday these places make murder illegal so people can’t kill others without violating the law.