Skip to content

Traces

Use Traces when you need the path of one request or a sampled view of slow and errored requests in scope.

Open Traces after choosing scope and time range. Use the trace list for aggregate search and the trace detail page for a single waterfall.

The Traces screen showing 24 hours of self-observability traces, latency percentiles, a scatter plot, and recent trace rows.

Common queries:

traces | where error_count > 0 | sort duration desc | limit 50
spans | where http.response.status_code >= 500 | stats count() by service, http.route
spans | where peer.service exists | stats count() by peer.service, bin(5m)

Trace detail shows span timing, service boundaries, status, attributes, resource attributes, events, links, and related logs when present.

Trace detail screen showing a waterfall, span metadata, related logs, and pivots to profiles or service views.

Traces require telemetry:query for the selected workspace or project.

  • Open logs filtered by trace_id or span_id.
  • Move from a slow route to metrics for the same service.
  • Move from a service in the waterfall to the Services page or Service Map.
  • Use exception or status fields as Explore predicates.
  • Open a linked profile when an OTLP profile sample names the trace.