Read a trace on Grafana Labs

One trace view, several ways to open it

Tue, 04 Aug 2026 18:58:09 -0400

One trace view, several ways to open it

By the end of the pipeline, your traces are stored and ready to open. But a stored trace only pays off once you can read it, and that’s the one skill the pipeline can’t hand you. Grafana can show a trace from more than one place, which is exactly what makes newcomers hesitate: pick one and you assume you’ve learned the wrong thing. You haven’t. They all open the same screen.

That screen is the trace view: the Grafana screen that shows a single trace. It lays the trace out as a timeline where each span is a row, stacked by when it started and nested under the span that triggered it, so you can see at a glance where the time went. That timeline is the span waterfall. Learn to read it once and the skill carries to every surface that shows a trace.

Where you open a trace

Explore: Grafana’s general-purpose screen for ad-hoc queries. It’s the traditional, universal way to open a trace.
Traces Drilldown: a visual way to explore traces without writing a query, so you can open a trace without a query box.

Both open the same trace view. You can also ask Grafana Assistant, Grafana’s built-in AI helper, to pull one up for you, for example “show me a slow checkout trace from the last hour”.

Whichever one you pick, you land in the same place:

Read a trace yourself

Tue, 04 Aug 2026 18:58:09 -0400

Read a trace yourself

You’ve seen how you reach a trace view. Now open a real trace and read its span waterfall yourself.

The hands-on path below runs on the public play.grafana.org demo, so you need no account and no setup. It lands you on a real trace with a ready-made query, then walks you through reading its span waterfall: the shape, the parent and child nesting, the durations, and the service and operation names. You open a span for its attributes, then compare a healthy request with a failed one to see how a trace’s shape alone can tell you what happened.

Learning path

Read a trace

Open a real trace with live demo data and learn to read its span waterfall (nesting, durations, and attributes) with no account or setup.

15 min

Beginner

Docs & blog posts

Open in Grafana Cloud

Complete this learning path directly in your Grafana Cloud stack, or in the Grafana Play stack, with an interactive learning experience.

Open in Grafana Cloud Try in Grafana Play

View on website

When you finish, you have the core skill, and it transfers to any trace on any surface. Two next steps build on that skill: a query language for finding specific traces on purpose, and connecting a trace to your other signals. The next two slides cover each one.

TraceQL: when you're ready for precision

Tue, 04 Aug 2026 18:58:09 -0400

TraceQL: when you’re ready for precision

Reading a trace someone found for you is one thing. Sometimes you need to go the other way and find every trace that matches a precise condition, for example “checkout requests slower than two seconds that returned an error.” Clicking through a visual explorer gets slow when your question is that specific.

For questions like that, Tempo has a query language built just for traces, called TraceQL. You describe the traces you want, and Tempo returns the ones that match, ready to open in the same trace view you already know how to read.

When to graduate to it

Start queryless: explore traces visually to answer most questions.
Reach for TraceQL: when you know exactly what you’re hunting for and want to pin it down precisely.

You don’t need it to read a trace, and this journey doesn’t teach its syntax. Just know it’s there for when your questions get sharper.

Correlation is powerful, but it isn't automatic

Tue, 04 Aug 2026 18:58:09 -0400

Correlation is powerful, but it isn’t automatic

A slow trace often raises a follow-up question: which metric alerted you to it, or what the logs for this exact request say. Jumping from one signal to another for the same request is called correlation, and it’s one of the most useful things you can do once you can read a trace.

These links run in both directions, and each one is its own clickable jump. A metric chart can carry a link that takes you straight to a representative trace behind a spike; that link is called an exemplar. In the same way, a log line can carry a link to the trace for that exact request, so you can open a log and jump straight to the trace that produced it.

It needs prerequisites

Correlation works once it’s set up, but the jumps don’t happen on their own. The jump from a log to its trace, for example, needs a few pieces in place first:

Trace IDs travel with your telemetry: your logs carry the trace ID of the request they belong to, so a log line can be tied back to its trace.
A link is configured: a Grafana data source setting, called a derived field, turns that trace ID in the log into a clickable link to the trace.
The target exists: the trace data source you’re linking to is connected in Grafana.

The diagram shows both jumps landing on the trace, with the log-to-trace jump resting on the three prerequisites above.

When a piece is missing, the jump quietly does nothing. So treat cross-signal correlation as a payoff you set up, not a switch that’s on by default. Wiring it up is part of the separate setup journey.

Resources

Tue, 04 Aug 2026 18:58:09 -0400

Resources

You can read a trace on your own. Where you go from here depends on what you want to do with that skill.

Continue learning

Pick the path that matches your goal:

Use Traces Drilldown to identify high latency service requests: Put trace reading to work investigating your own traces in Grafana Cloud.
Send traces to Grafana Cloud using Alloy: Get your own traces flowing, from instrumenting your application to sending spans with Alloy.
Visualize traces in a Grafana Cloud dashboard: Build dashboards from your traces and start writing TraceQL.

Documentation

Introduction to tracing: Revisit traces, spans, and the pipeline.
Trace structure: A closer look at spans, nesting, and attributes.
Traces Drilldown: The queryless way to explore your traces.
TraceQL cookbook: Ready-to-use TraceQL query examples for when you’re ready for precision.

Community

Grafana Community Forums: Ask questions and share what you’re building.
Grafana Slack: Join the conversation with other Grafana users.