Show HN: ML condenses billions of logs into a tiny snapshot your LLM can debug

Category: observability

Tags: log-clustering, anomaly-detection, self-hosted, observability, ml, opentelemetry

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

Rocketgraph is a self-hosted log clustering and anomaly detection tool that uses deterministic ML algorithms to condense millions of logs into a few structural templates, flagging anomalies without any LLM or external data leaving your VPC. Its innovative combination of Drain3, Isolation Forest, and Half-Space-Trees for real-time, reproducible log analysis, alongside an OpenTelemetry auto-instrumentation agent, fills a clear gap in existing observability stacks by highlighting what's unusual rather than just what you searched for.

Target audience: devops, backend devs, site reliability engineers (SREs), data engineers

Repository: https://github.com/Rocketgraph/rocketgraph · Python · NOASSERTION · 138 stars

View on Hacker News