The environment project made its own plan obsolete, and the team let it

The plan was reasonable when it was written. Build an environment logger on Home Assistant, feed it room sensors and machine telemetry, and have one place to look at how hot the room gets and why. Work started.

Then, partway through, someone noticed that the machine half of it already existed. BMC temperatures, fan speeds, PDU and UPS wattages had all arrived in the fleet’s Prometheus stack on their own, as a side effect of unrelated monitoring work finished earlier, with alerting and dashboards already built around them. The logger under construction was going to be a second, worse copy of a timeline that was already live.

Finishing it would have been the easy choice. The work was underway, nobody had told the team to stop, and the result would have functioned. Instead the architecture was inverted. Prometheus became the single timeline for both machine and room telemetry, and Home Assistant’s role was narrowed to the two things it uniquely does well: ingesting sensors and driving actuators. Derived numbers and alerting live in the monitoring stack. Home Assistant’s own sensors export outward into the same scrape job as everything else, which is the part that was checked live rather than assumed.

The resulting design is cleaner than either original plan, and it exists only because someone was willing to say the plan needed to change rather than protect work already spent. The transferable habit is to notice when a project has gone obsolete mid-build, and to treat the sunk work as sunk. The question to ask at the halfway point is this: if we were starting today, knowing what now exists, would we build this at all.