Skip to content

Traces#

This page shows you how to send distributed traces from your application to KubeStack+ and explore them in Grafana.

How it works#

Stakater operates Tempo for trace storage and Grafana for visualization. Your application emits spans through the OpenTelemetry SDK; the SDK exports OTLP to the in-cluster collector; the collector routes them to Tempo; you query them in Grafana through Forecastle. Trace context is propagated across services so a single request shows up as one end-to-end trace.

What you do#

Add the OpenTelemetry SDK or auto-instrumentation agent for your language to your application, then set the OTel SDK environment variables on your pod via deployment.env in your values.yaml for the Stakater Application Helm Chart:

deployment:
  env:
    - name: OTEL_SERVICE_NAME
      value: your-service-name
    - name: OTEL_EXPORTER_OTLP_ENDPOINT
      value: http://otel-collector.observability.svc.cluster.local:4317
    - name: OTEL_TRACES_EXPORTER
      value: otlp

For Java, Python, Node.js, and .NET, the auto-instrumentation agent captures web request, database, and messaging spans without code changes. Open Grafana from Forecastle, switch to Explore, select the Tempo data source, and search by trace ID, service name, or duration. From a trace you can jump to the matching logs in Loki or metrics in Mimir.

Next step#

Continue to Dashboards to build a Grafana dashboard that combines metrics, logs, and traces.