What is avuru obs?
avuru obs is an open-source, all-in-one observability platform. Traces, metrics, logs, continuous profiling — and per-service energy & carbon — in a single install, on one storage engine, behind one UI. It consolidates a full multi-service observability stack into a much smaller footprint, with no SDK rollout.
The wedge is simple:
A fresh Kubernetes cluster goes from zero to a live service map in under five minutes — with zero changes to your applications.
Why it's different
- eBPF auto-discovery. The
sensorruns upstream OBI (OpenTelemetry eBPF Instrumentation) for zero-code traces, and your service map is derived from those trace spans. OBI's built-innetworkfeature fills in un-instrumented edges. No SDKs, no sidecars, no YAML archaeology — install the chart and watch the map light up. - One engine. Every signal lands in ClickHouse. One thing to size, back up and reason about — not four.
- OpenTelemetry-native. Already instrumented? Point your OTLP exporters at
the gateway (
:4318/:4317) and you're done. avuru obs is a drop-in OTLP backend. - Low footprint. A minimal collector distro and a single Go control plane — not a constellation of services.
- Green IT built in. The opt-in green module attributes Wh and gCO₂e per service (CNCF Kepler, RAPL) — carbon budgets and a CSRD-ready export, computed on your cluster with zero instrumentation and no data egress.
The five signals
| Signal | How it arrives | Where it lives | Status |
|---|---|---|---|
| Traces | eBPF + OTLP | ClickHouse | Available |
| Logs | collected, correlated by trace_id | ClickHouse | Available |
| Metrics | RED + OTLP | ClickHouse | Available |
| Profiling | continuous (eBPF profiler) | ClickHouse | Available · opt-in |
| Energy | CNCF Kepler (RAPL) | ClickHouse | Available · opt-in |
:::tip What works today Traces, logs, metrics, profiling, the service map, error tracking, service health, alerting and the green module (energy & carbon, opt-in) are live. See Feature status for the full picture. :::
The pieces
avuru obs is a handful of components, each doing one job:
sensor— eBPF DaemonSet (flows, traces, RED metrics, logs, profiles, energy via Kepler — opt-in).gateway— a minimal OpenTelemetry Collector distribution.hub— a single Go binary: REST/WebSocket API + OpAMP config plane.ui— a static SPA served from its own pod.- ClickHouse — the one storage engine for all signals.
See Architecture for the full data flow.
:::tip Ready in 30 seconds?
Jump straight to the 30-second guide — the
fastest path from helm install to your first trace.
:::
Where to go next
- 30 seconds — install and see the map.
- Core concepts — signals, context, propagation.
- Your first trace — verify end-to-end.
- Integrations — languages, frameworks, databases, infra.
- Compare — how avuru obs stacks up against Jaeger, the Grafana stack, Prometheus and Datadog, and how to migrate.
:::note Status avuru obs is at v0.2 and under active development. APIs and schemas may change between minor versions. Licensed under AGPL-3.0. :::