I’m a long-time time series nerd, I’ve worked with InfluxDB, TimescaleDB, ClickHouse, and more, across everything from monitoring fleets to tracking medical devices. But recently, I started exploring RedisTimeSeries again… and I was surprised by how much the Redis Stack has evolved.<p>Between RedisTimeSeries, RedisJSON, RediSearch, and Streams, I realized: this could actually be the backbone for a full observability stack.<p>So I built rtcollector, a modular, Redis-native observability agent. It’s written in Python, configured with YAML, and designed to push system, container, and database metrics into RedisTimeSeries with labels and retention. Think of it as a Telegraf alternative, but for Redis.<p>Right now, I’ve implemented input plugins for:
• Linux: CPU, memory, disk, I/O, network
• macOS: CPU, memory, disk, I/O, network
• Docker: container stats via API
• Redis, MySQL, PostgreSQL<p>The idea is to keep it simple, extensible, and Redis-first.<p>Next steps:
• Native logs via RedisJSON + RediSearch (already prototyped!)
• Support for Redis Streams (for traces/events)
• Dashboards in Grafana using the Redis data source<p>If you’re into observability, Redis, or just like building small purposeful tools, I’d love your thoughts or contributions. It’s early, but already useful for homelabs, edge boxes, and anyone tired of deploying 10 containers just to get CPU metrics.<p>Repo: <a href="https://github.com/xe-nvdk/rtcollector">https://github.com/xe-nvdk/rtcollector</a>