When you open Pyroscope and see a flame graph for your Go application, symbolization is what makes it possible. Here’s a deep dive on how that process works with eBPF profilers.

Grafana is an open-source analytics and monitoring platform that allows users to visualize and analyze metrics from various data sources. The Grafana blog covers topics such as data visualization, monitoring best practices, and new features in Grafana software. Developers and data engineers can learn how to leverage Grafana for monitoring infrastructure, analyzing performance metrics, and creating interactive dashboards for data-driven decision-making.

Grafana Labs

A detailed walkthrough of how the OpenTelemetry eBPF profiler converts raw memory addresses into readable function names for Go programs. The post explains the symbolization pipeline step by step: finding the binary via /proc/maps, loading the .gopclntab section, performing binary search to resolve addresses, and caching results. Key insight: Go embeds .gopclntab in every binary (even stripped ones) because the runtime needs it for panic traces, making Go uniquely suited for eBPF profiling. The post covers performance optimizations (O(log n) binary search, LRU frame cache achieving sub-microsecond lookups), failure scenarios like CGO and inlining, and practical diagnostic commands (readelf, nm, file) to inspect binaries.

From raw data to flame graphs: A deep dive into how the OpenTelemetry eBPF profiler symbolizes Go