Deployed Immich v3, uploaded photos, verified the ML search works ("sunset over water" ranked correctly, 2.2s on CPU). Real numbers: 2.4GB RAM working set, 4.7GB images, per-container breakdown. Tagged with selfhosted, docker, photos, opensource. Originally published on FreeTierWatch, where we actually deploy and measure…
WireGuard key rotation usually gets planned the day after a laptop disappears from a taxi, when nobody can say which of the 40 peers on the hub belongs to it. The tunnel itself is simple. The operational picture around it (who…
Series: Part 1: choosing and creating the server · Part 2: hardening and configuring it · Part 3: migrating data and cutting over I recently moved a small production API from one VPS provider to Hetzner Cloud. The stack is ordinary: a Rust web server,…
I spent a lot of time on the short video cutting down 2 hours of content into 5 minutes, so this is brief. Since my last post, I've setup my Juniper switch behind my consumer router and handled the double NAT using…
A lot of one-person setups never make that leap. They run a model inline in a Jupyter process, or start a server by hand when they need it. That works until it does not. Here is the workflow I landed…
This is Meisei Rising. “I am worried about putting internal company documents into external AI services.” One option when you feel this way is “local AI,” which runs generative AI on your company PC. A free software widely used as…
Your laptop needs no GPU and no model file. The model runs on the server; your laptop sends text over your network. By default both servers accept calls only from their own computer. Ollama on a Linux server runs as…
💡Note: This guide works perfectly with the latest version of Pi-Hole v6.5 – Docker Version 2026.10.0 and UGOS PRO 1.20.0.0142 – 1.91.10.0022 💡Note: Check out my new guide on how to Install Pi-Hole V6 on Your Synology NAS. Please Support My work by Making a Donation. ⚠️Mandatory…
If you maintain a monorepo with Nx or Lerna, you know how much remote caching helps: whatever a developer or a CI job has already built doesn't need to be built again by anyone else. The problem is where to…
Survival LLM, the offline AI assistant Hacker News spent this week arguing about, got 14 of 22 blackout and first-aid questions right on a 16 GB machine. The other eight missed the main point of the official guidance, and these were emergency questions.
App catalogs, reverse proxies, SSL, DNS, and best practices for self‑hosting at home.
An article joins this feed when it matches these terms. Each one is also a search of its own.