Skip to Content
ModulesTraces

Traces

The Traces module provides distributed tracing for understanding request flow across microservices. It captures end-to-end latency, identifies bottlenecks, and correlates trace spans with logs and metrics.

Key Features

  • End-to-end trace visualization — Waterfall view (timeline by span) and optional Flame graph view (nested spans by depth and duration) when span data is stored for the trace
  • Flame graph API — GET /api/traces/{trace_id}/flame returns { data: { trace_id, total_duration_ms, flame_spans[] } } with start_ms relative to trace start, depth, service, status, and span_id for selection
  • Service-level latency — Aggregate latency percentiles (p50, p95, p99) per service and endpoint
  • Error tracking — Surface spans with errors and link them to the originating service
  • Trace-to-log correlation — Click any span to see logs emitted during that span’s execution
  • Trace-to-metric correlation — View infrastructure metrics (CPU, memory) for the host/container during the trace
  • Sampling controls — Configure head-based or tail-based sampling rates per service
  • OpenTelemetry native — Ingest traces via OTLP, Jaeger, or Zipkin protocols
  • Critical path analysis — Highlight the spans that contribute most to total request latency

How to Access

Navigate to Traces in the left sidebar. The default view shows a trace search interface with service and latency filters.

Basic Usage

  1. Open Traces from the sidebar (under OBSERVE)
  2. Filter traces by service, operation, minimum duration, or status (OK / Error)
  3. Click a trace row to open the detail drawer
  4. Switch View between Waterfall and Flame graph — waterfall shows a Gantt-style timeline; flame graph stacks spans by depth for quick hotspot scanning
  5. In flame view, click a block to focus span details (when span_id is present)
  6. Examine individual spans — click for attributes, error stack traces, and links to Logs and Services
  7. Use View logs for this trace from the drawer header to correlate with logs

Some roadmap items (compare two traces, latency alerts UI) may vary by deployment; use Metrics and SLO modules for alerting where integrated.

LogsNetFlow & traffic