Exploring APM data

APM data lives under the APM section of the sidebar.

Services

APM → Services lists one row per service.name, with:

Column	Meaning
Requests, req/min	Number of entry spans: server and consumer spans, plus root spans of background jobs. Outgoing client and producer calls aren't counted.
Error rate	Share of requests with status ERROR, or HTTP 5xx on server spans with no status set.
Avg, p50, p95, p99	Request latency.
Spans	All spans of the service, including outgoing calls and internal work.

Service page

Click a service to see its requests, errors and latency (p50, p95, p99) over time, and a table of its operations (span names such as GET /orders/:id), each with its own numbers. Click an operation to open its traces.

Traces

APM → Traces lists one row per request, newest first.

• Side facets: service, status (ok or error), operation, HTTP status class (2xx, 4xx, 5xx) and environment. You can select several values in a facet. Values within a facet are combined with OR; different facets are combined with AND.
• Duration: quick presets for slow requests, for example over 100 ms, 500 ms or 1 s.
• Search: an operation name, route or a full trace ID.
• Time range: at the top of the page.

Trace view

Clicking a trace opens every span of the request across all services.

Flame graph (default)

• Each bar is a span, colored by service. Children sit under their parent.
• Spans with errors are marked in red.
• Hover a span for its duration, start offset and share of the whole trace.
• Double-click a span, or choose Zoom to span, to zoom in on it. Reset zoom shows the whole trace again.

Waterfall

Switch to Waterfall for one row per span on a timeline. This view suits long traces and helps you find where time goes between calls.

Span details

Select a span to see:

• Attributes, such as HTTP method, route, status, database statement and peer service.
• Resource: service, host, environment, SDK version.
• Events, for example exceptions with their stack traces.
• Logs that carry the same trace ID.

OTel metrics

APM → OTel metrics shows every metric your apps export, such as HTTP server duration, runtime memory and custom counters.

1. Pick a metric in the side panel.
2. Filter by service.
3. Compare the series on the chart.

How values are shown:

Metric type	Shown as
Counter (monotonic sum)	Rate per second
Histogram / summary	Average observed value (Δsum ÷ Δcount)
Gauge / up-down counter	Average value

A counter needs two data points before a rate can be drawn, so it appears after the second export.

Logs from OpenTelemetry

OTel log records appear in Logs together with the agent's file logs:

Log field	Comes from
service	service.name
host	host.name (or k8s.pod.name, service.instance.id, then otel)
status	severity: 1–8 debug, 9–12 info, 13–16 warning, 17–20 error, 21–24 critical
tags	source:otel and env:<deployment.environment>

Logs with a trace ID show a View trace link that opens the trace view.