Hey — Josh here.
If you subscribed at buildahomelab.dev and this is the first email you've ever gotten from me, that's because it is the first email I've ever sent. Welcome to issue #1.
The news: the site now lives at peira.dev, under the name Peira Labs (say PAY-rah — it's the ancient Greek word for "trial" or "attempt," which is the most honest description of a homelab I've found). Everything you've bookmarked still works; the old domain redirects permanently. Same lab, same posts, same me — just one name over all of it. And that's why this newsletter is called Trial & Error: it's what peira means, and it's how everything here gets built.
The field note: the rename itself handed me this issue's war story. Mid-cutover, I deployed the site's config with the new domain in it — and the old domain went dark. Not stale, not redirecting: gone. It turns out my hosting tool treats the config as the complete list of domains that should exist, so any domain you leave out doesn't just get ignored — it gets detached, DNS record and all. The fix was quick, but the lesson stuck: when a tool reconciles state instead of adding to it, everything you don't mention is something you've asked it to remove. The same rule that makes infrastructure-as-code trustworthy makes a half-written config destructive.
From the last two weeks:
Turn a Hand-Built Proxmox Cluster Into Code Without Breaking It — the start of a five-part series on importing a live, hand-built cluster into OpenTofu and Ansible with zero changes. If you only read one, read this one.
Rebuild Your Homelab From Zero: What Survives a Total Loss — the series finale: what code restores, what backups restore, and the survival inventory that decides whether a rebuild is a bad afternoon or a bad month.
RAG for Your Homelab Wiki: Answers That Cite Their Source — asking your own documentation questions and getting answers that cite page and heading, with a staleness warning built in.
Homelab Capacity Planning: What If a Node Dies Tonight? — a little simulator that answers the question before the failure does.
The Docker localhost Trap: 24 Hours of Red, Zero Real Downtime — a wall of red monitoring alerts, none of them real, all of them my fault.
Lab status: four Proxmox nodes, all green. The big recent milestone was backups: three tiers now, with the offsite copy finally in place. Next up is the thing every backup post warns you about — actually restoring one on purpose.
New here or catching up? Start Here is the guided path through everything, in order.
See you in two weeks,
Josh