go error: go tool pprof - profile contains no samples - how to fix

Not specified by the original author.

go error: go tool pprof - profile contains no samples - how to fix

TL;DR

An empty profile means the profiler never got to sample: the process exited before sampling started, signals are blocked, or the profile window captured nothing. Keep the process alive through the whole window, make sure SIGPROF is not blocked, and give it at least 30 seconds of real work to sample.

go tool pprof - profile contains no samples

Steps

1. Make sure the process outlives the profile window: run the profile for 30s against a process that stays up and does work the whole time. Expected: A longer-lived process produces samples.

2. Check SIGPROF is not blocked or ignored by the app or its container runtime. Some sandboxes swallow profiling signals. Expected: Signals reach the process (test with a trivial CPU profile of a busy loop).

3. If profiling a short-lived command, use -cpuprofile to write the file instead of the HTTP endpoint, and verify the file is nonzero before opening it. Expected: The output file has real size.

4. Confirm the binary was built with the same Go version and not stripped of symbols in a way that breaks sampling (it still samples, but verify the basics first). Expected: pprof opens the profile and shows functions.

5. For HTTP-based profiles, hit /debug/pprof/profile?seconds=30 directly and watch for errors in the app log during the window. Expected: The endpoint returns a valid profile instead of an empty one.

When to use

• go tool pprof reports no samples
• the profile file is empty or tiny
• CPU profiling a short-lived process

When not to use

• pprof shows samples but symbols are missing (build with symbols; different issue)
• the profile is for heap or goroutines (those do not need sampling windows)
• the process is idle (an idle process legitimately has no CPU samples)

Tool compatibility

• Go 1.19+
• runtime/pprof, net/http/pprof

Variant phrasings

• "pprof profile contains no samples"
• "go tool pprof empty profile"
• "pprof no samples collected"

Why it happens

CPU profiling works by interrupting the process with SIGPROF on a timer and recording the stack. If the process dies before the first tick, blocks the signal, or does no CPU work during the window, there is nothing to record and pprof reports an empty profile.

Edge cases

• profiling inside some containers needs the SYS_PTRACE-adjacent permissions; without them the signal path can fail silently
• a program sleeping in syscalls the whole window yields no CPU samples; that is correct, not a bug
• mixing -cpuprofile output with the HTTP endpoint in one run can truncate the file; use one method per run

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_8qoTVTlzPnmbqQFxt14qbA

Resolved from the public thread: https://vectle.com/posts/pst_8qoTVTlzPnmbqQFxt14qbA

Thread query: go error: go tool pprof - profile contains no samples - how to fix

TL;DR: An empty profile means the profiler never got to sample: the process exited before sampling started, signals are blocked, or the profile window captured nothing. Keep the process alive through the whole window, make sure SIGPROF is not blocked, and give it at least 30 seconds of real work to sample.

Not specified by the original author.

TL;DR

Use 0, not None, on services that sit mid-trace. - Profiles sample relative to traces, so changing the trace rate silently changes profile volume too. - Spike protection drops excess volume per project regardless of your rates; sampling strategy and spike protection are complementary, not substitutes.

Steps

1. One flat sample rate is a compromise that pleases nobody: too high for hot paths, too low for rare-but-critical ones. Sample by value instead.
2. After a week, compare per-route sample counts against traffic: hot paths cheap, critical paths complete, baseline representative. Adjust the sampler when routes change; a renamed endpoint silently falls back to the baseline.

When to use

You are seeing this: One flat sample rate is a compromise that pleases nobody: too high for hot paths, too low for rare-but-critical ones. Use this skill when you run into "Workflow: trace sampling as a strategy".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

Versions mentioned in the source: 0.05-0.2. If you are on something much newer or older, the details may have shifted.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

Researched against live Sentry docs 2026-09-26. The Python options page (docs.sentry.io/platforms/python/configuration/options/) documents tracessamplerate, tracessampler, the 0-vs-None semantics (0 keeps continuing incoming traces, None disables tracing), and profilessample_rate. The quotas page (docs.sentry.io/pricing/quotas/) documents per-category quotas and spike protection. The Laravel guide warns a trace rate of 1.0 burns quota fast.

Not specified by the original author.

TL;DR

Set the trace rate first, then choose the profile fraction of it. Check the profiling quota category in usage stats.

Steps

1. profiles_sample_rate is relative to traces_sample_rate, not absolute. If traces sample at 0.1 and profiles at 0.5, only 5% of requests produce profiles. Set the trace rate first, then choose the profile fraction of it.

Sentry.init({
 dsn: "[your-dsn]",
 tracesSampleRate: 0.1,
 profilesSampleRate: 0.5, // fraction of sampled traces
});

2. Start low: 0.1 traces with 0.5 profiles gives 5% profile coverage, plenty for hotspot detection.
3. A known-hot endpoint shows its expensive functions in the profile summary, quota usage grows linearly with the rollout (no cliff), and no service exceeds its share. Revisit rates quarterly: as code changes, yesterday's hotspots cool and new ones appear.

When to use

You are seeing this: Profiling shows which functions burn CPU, but it samples on top of your traces. Use this skill when you run into "Workflow: roll out continuous profiling without quota shock".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

No specific versions are mentioned in the source material, so treat the fix as generally applicable and check the examples against whatever you have installed.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

Researched against live Sentry docs 2026-09-26. The JS profiling page (docs.sentry.io/platforms/javascript/profiling/) documents the profiling integration and profilessamplerate. The Ruby guide (docs.sentry.io/platforms/ruby/guides/rails/) documents profilessamplerate as relative to tracessamplerate. The Python options page documents the tracing options model. The quotas page (docs.sentry.io/pricing/quotas/) documents per-category quotas.