;
};
`Python` · `IoT` · `CrateDB` · `Docker` · `Apache Superset`
## Overview
The data source is a TP-Link Tapo H200 hub with T315 temperature and humidity sensors. A Python collector starts up, auto-discovers all T315 sensors registered under the H200, and polls them roughly every 60 seconds, writing readings to a CrateDB time-series database. Apache Superset provides dashboards. The whole stack is packaged as Docker Compose and runs on a Synology NAS or any Docker host.
The H200 hub API gives no online/offline status for child sensors: after a sensor drops, the hub keeps returning the last cached reading. The system cross-checks four signals to detect stale data and writes NULL for temperature and humidity when at least N checks fail, logging the event to `stale.log`.
## Core features
* Auto-discovers all T315 sensors under the H200 on startup, no manual device list required
* Collects temperature, humidity, battery level, and RSSI every \~60 seconds
* Four-signal cross-validation (RSSI threshold, RSSI freeze, simultaneous temperature and humidity freeze, device\_time freeze): stale only when at least N checks fail (default N=2), avoiding single-signal false positives
* Stale readings write NULL to temperature and humidity columns, preserving `is_valid` and `stale_reasons`; events are also written to `stale.log`
* Apache Superset dashboards with pre-built per-minute, hourly, daily, weekly, and monthly aggregate views
* Optional Cloudflare Tunnel for remote access without opening inbound ports
* Automatic reconnect and retry on failure, `restart: unless-stopped`
## Architecture
## Quick start
```bash theme={null}
cp .env.example .env
```
Fill in the Tapo account credentials and the H200 LAN IP address.
```bash theme={null}
cd docker
docker compose -f docker-compose.prod.yml up -d --build
```
```bash theme={null}
docker compose -f docker-compose.prod.yml --profile init up superset-init
```
Open `http://NAS-IP:4200` in a browser and run `scripts/init_db.sql` in the SQL console. The service does not create tables automatically.
```bash theme={null}
docker compose -f docker-compose.prod.yml logs -f climate-monitor
```
Sensor names and readings appearing in the log confirm the collector is active.
## Stale-data validation
The H200 API does not expose child device connection state, and the hub continues to return cached values after a sensor disconnects. The system cross-checks four signals:
A reading is marked stale only when at least `VALIDATOR_MIN_FAILED_CHECKS` (default 2) checks fail simultaneously. The freeze window is computed from `frozen_window / collection_interval`.
## Notes
`.env` contains Tapo account credentials; keep it out of version control. Local development (`docker-compose.pc.yml`) and production (`docker-compose.prod.yml`) use different bind-mount paths. Run `scripts/init_db.sql` manually after the first deployment; the service does not auto-create tables.
## In practice
Suitable for any long-term, low-cost multi-point temperature and humidity monitoring: lab environment logging, server room baseline monitoring, home climate management. The stale validation gives NULL values a precise meaning (sensor offline) rather than silently persisting incorrect data, making it straightforward to exclude invalid intervals in downstream analysis.
## Links
* GitHub: [felimet/climate-monitor](https://github.com/felimet/climate-monitor)
* CrateDB docs: [cratedb.com/docs](https://cratedb.com/docs)
* Apache Superset docs: [superset.apache.org/docs](https://superset.apache.org/docs/intro)