This is what got me started. I had a drive throwing write errors ~ pulled the SMART data and there were a pile of indicators that had been quietly getting worse: reallocated sectors climbing, uncorrectable errors, pending sectors. The driveās overall SMART status still said āPASSED.ā That was the moment I realized I needed something that watches the actual early warning attributes, not just the pass/fail flag.
So I built SMART Sniffer.
Itās a HACS integration + lightweight Go agent that runs on each machine. The agent wraps smartctl and serves data over HTTP ~ HA discovers it via mDNS. Each drive gets its own device with sensors, and thereās an āAttention Neededā sensor that specifically monitors the attributes that predict failure. When something looks off, it fires persistent HA notifications automatically ~ no automations or blueprints needed.
For anyone still using the cron + JSON file approach: it works fine for local drives, but if you want multi-machine monitoring that catches the warning signs your drives wonāt tell you about, might be worth a look.