Crossroads

The etcd cluster is up and waiting for data. Here’s why copying everything over as-is didn’t work out — and how the migration turned into a full rethink of the data schemas.

Breaking down what was wrong with the old architecture and how I redesigned it: from a tree to two independent clusters, zero trust, one domain per cluster, and Docker-based onboarding.

What KV stores are, why they matter for distributed systems, and how etcd became the foundation of operational data storage in Sigil Gate.

I’ve been using cron my whole life. Turns out there’s been something better for quite a while.