Effortless, Standardised Homelab Observability With eBPF
summary
Home lab gets unified observability by running Beyla (eBPF → OpenTelemetry) plus Prometheus and Grafana, which auto-generates HTTP request/error/latency metrics and traces across languages without restarting pods. On Raspberry Pi a kernel must be rebuilt for BTF, and eBPF misses in-process attributes (customer IDs) and full SQL query sanitation, so native OpenTelemetry instrumentation still gives higher-fidelity attributes but costs manual context propagation work. Practical wins: one Helm install instruments dozens of apps; add Blackbox exporter probes for availability and keep eBPF as low-effort, lower-detail coverage.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:10Home lab monitoring challenges
- 7:01eBPF and OpenTelemetry fit
- 10:32Installing Baila and kernel config
- 12:20eBPF instrumentation under the hood
- 15:48Setting up SLOs and alerts
- 18:16Push vs pull monitoring
- 20:24Manual OpenTelemetry instrumentation
- 24:39Manual vs eBPF instrumentation
- 26:38Q&A
related talks
31mEffortless, standardised homelab observability with eBPF
34mNetwork Observability with eBPF and OpenTelemetry
31meBPF for Observability: The Good, the Bad, and the Ugly
30mAgentless instrumentation with eBPF, Prometheus and OpenTelemetry
37meBPF: Revolutionizing Observability and Telemetry for Modern Applications
31mZero instrumentation observability with eBPF
35mAn Exemplary Path: Leveraging EBPFs and OpenTelemetry To Auto-instrument for Exemplars
27mStep By Step Kubernetes Observability with eBPF