Skip to content

Build your first dashboard

Use this guide when you want one page that shows the health of your services at a glance, to keep open on a screen or to share with your team.

You will build the dashboard below from trace data: how much each service does, how often it fails, and how slow it is. No metrics setup is needed.

The finished HotROD overview dashboard: a Spans per service line chart, an Errors per service bar chart and a P99 latency by service table.

You need the Editor or Admin role.

  1. Open Dashboards in the sidebar and click New dashboard.
  2. On the Blank tab, replace Sample Dashboard in Title with a name, such as HotROD overview, and add an optional Description. In Tags, type team:demo and press Enter.
  3. Click Create dashboard. The new dashboard opens on Welcome to your new dashboard.

An empty dashboard: Welcome to your new dashboard, with a Configure your new dashboard step and an Add panels step with New Panel.

A panel is one chart on the dashboard.

  1. Click New Panel. In the New panel picker, click Time Series, then Add panel. The panel editor opens: the query on the left, a preview of the chart in the middle and the panel settings on the right.
  2. In the query card, change Source from Metrics to Traces. The aggregation is count(): the number of spans.
  3. Open Result, click Group By and pick service.name, to draw one line per service.
  4. In Legend format, enter {{service.name}}, so each line is labelled with its service name.
  5. On the right, set Title to Spans per service.
  6. Click Run Query to check the chart, then Save changes.

The panel editor: a Traces count() query grouped by service.name with a legend format that names each line after its service on the left, the Spans per service chart in the middle, and Panel Details on the right.

  1. Click New Panel at the top of the dashboard, pick Bar Chart and click Add panel.

  2. Change Source to Traces and type this filter in Where:

    has_error = true
  3. In Result, group by service.name and set the legend to {{service.name}}. Set the title to Errors per service.

  4. Click Run Query, then Save changes.

  1. Click New Panel, pick Table and click Add panel.
  2. Change Source to Traces. Replace count() with p99(duration_nano), the duration that 99 in 100 spans stay under.
  3. In Result, group by service.name. Set the title to P99 latency by service.
  4. Durations are stored in nanoseconds. On the right, open Formatting & Units, and under Column units pick Nanoseconds for the p99(duration_nano) column. The table now shows values such as 737.86 ms.
  5. Click Run Query, then Save changes.
  • Drag a panel by its header to move it, and drag its bottom-right corner to resize it. The layout is saved when you let go.
  • Change the time range at the top of the page; every panel follows it.
  • Click Share (the globe icon at the top right), then Copy page link. Anyone with access to FuseScope who opens it sees the same dashboard and time range.
  • To group panels later, use Actions › New section. See Sections.

A dashboard with three panels that update with the time range. Open it from Dashboards at any time; click the pin icon on its row to find it in the Pinned view.