Go 1.25 Introduces Flight Recorder for Real-Time Execution Tracing

Go 1.25 Introduces Flight Recorder for Real-Time Execution Tracing

September 26, 2025 — The Go team announced today that Go 1.25 includes a built-in flight recorder for runtime execution traces, giving developers a powerful new way to diagnose production issues without pre-planning or massive data collection.

The flight recorder continuously buffers the last few seconds of a Go program's execution trace in memory. When a problem is detected—such as a timeout or health check failure—developers can request a snapshot of exactly the problematic window, rather than capturing an entire trace or relying on random sampling.

“This is like having a black box for your Go services,” said Carlos Amedee, a Go team engineer. “You don't need to start recording ahead of time; the recorder is always ready, and you grab the tape after the incident.”

A Quick Look at Execution Traces

Go execution traces are detailed logs produced by the runtime that capture goroutine scheduling, system calls, and other low-level events. They help developers understand why goroutines are executing or, crucially, why they are not—making them essential for debugging latency issues.

Previously, collecting a trace required explicit Start and Stop calls via the runtime/trace package. This worked fine for tests or short-lived tools, but long-running web services—Go's bread and butter—could not afford to record all activity.

“A web server might be up for weeks,” said Michael Knyszek, also on the Go team. “Collecting an end-to-end trace would produce terabytes of data that is mostly noise. The flight recorder changes that dynamic completely.”

Background: The Problem with Pre-Planned Tracing

Before Go 1.25, engineers had two main options for tracing in production: continuous collection or random sampling. Both come with significant drawbacks.

• Continuous collection requires infrastructure to store, triage, and process large volumes of data—most of which contains nothing interesting.
• Random sampling can catch anomalies but demands heavy data pipelines and may miss rare, critical events.

Neither approach helps when a specific issue (like a request timeout) occurs without warning. By the time a problem is noticed, it is already too late to start tracing. The flight recorder solves this by maintaining a rolling buffer that can be frozen on demand.

What This Means for Go Developers

The flight recorder is a precision diagnostic tool. “It's like a scalpel cutting directly to the problem area,” Knyszek explained. “Instead of drowning in data, you get exactly the seconds that matter.”

This feature is particularly valuable for microservices and web applications where latency spikes or failures are transient. Developers can now add a simple API call to snapshot the trace when error conditions are met, dramatically reducing mean time to resolution (MTTR).

Go 1.25 also improves the underlying execution tracer performance, so enabling flight recording has minimal overhead. The buffer size is configurable, allowing teams to balance memory usage against trace depth.

The flight recorder is available starting with Go 1.25. For more details, see the official documentation and the Go Blog post.