Skip to content

Service catalog

Service Catalog provides a per-service metrics table built from your distributed traces. It summarizes request volume, error behaviour, and latency for every service in the selected trace stream so that you can compare services side by side and decide where to investigate next.

Key points:

  • It gives you a tabular, sortable view of service health across one trace stream.
  • It highlights services with elevated error rates and slow latency so you can spot problems quickly.
  • It complements Service Graph, which shows the same kind of health information as a visual topology.
  • From any row you can jump straight into Traces filtered to that service for detailed debugging.

Service Catalog reads recent trace activity for the selected stream and time range. If there is no trace data, the view displays a message indicating that no services were found.

Where to find this

  1. Sign in to OpenObserve.
  2. Select Traces in the left navigation panel, then choose Service Catalog from the Spans | Traces | Service Graph | Service Catalog toggle in the search bar.

The catalog loads automatically for the selected trace stream and time range. The Service Graph option in the same toggle is Enterprise-only, but Service Catalog is available in all editions.

What service catalog displays

Service Catalog displays one row per service discovered in the selected trace stream. Each row shows a health status, request and error counts, error rate, and latency percentiles and durations.

Columns

The table includes the following columns.

Column Description
Service The service name discovered from the trace stream.
Status A health status pill derived from the error rate. See Health status.
Requests Total number of spans (requests) recorded for the service in the selected range.
Errors Number of spans with an error status.
Error Rate Percentage of requests that ended in error. The value is color-coded as the rate increases.
P50 Median (50th percentile) span duration.
P95 95th percentile span duration.
P99 99th percentile span duration. Highlighted in orange when it exceeds 1 second.
Avg Duration Average span duration for the service.
Max Duration Longest span duration recorded for the service.

You can sort the table by any column. By default, services are sorted by health status, with the least healthy services first.

Operations

Per-operation metrics are not shown as a column in the table. To see operations for a service, open the service detail side panel by clicking a row, then use the Operations tab.

Health status

Each service is assigned a health status based on its error rate. The status appears as a colored pill in the Status column and is summarized in pills above the table.

Status Error rate
Healthy Less than 1%
Degraded 1–5%
Warning 5–10%
Critical Greater than 10%

In addition to the status pill, latency is flagged independently: a P99 value greater than 1 second is highlighted in orange to draw attention to slow services even when their error rate is low.

Data source

Service Catalog uses distributed traces as its single data source. For the selected trace stream, OpenObserve aggregates spans by service to compute request counts, error counts, error rate, average and maximum duration, and the P50, P95, and P99 latency percentiles.

Service Catalog does not use logs or metrics directly.

Controls

Stream selector

Stream selector

Select the trace stream to build the catalog from. The catalog is computed for a single stream at a time. Your selection is remembered for future visits.

Service-name filter

Service-name filter

Enter text in the filter field to narrow the table to services whose name contains your input. The match is case-insensitive, and a counter shows how many of the total services match.

Date-time picker

Date-time picker

Use the date-time picker to set the time range for the metrics. You can choose a relative range or an absolute start and end time. Changing the range reloads the catalog.

Refresh

Refresh

Use the refresh control to reload the catalog on demand with the current stream and time range.

Drill-down

Service Catalog is a starting point for investigation. From a service row you can move directly into detailed trace analysis.

  • Click a service row to open the service detail side panel, which includes Operations and related tabs for that service.
  • Select View Traces to open the Traces view filtered to that service. OpenObserve sets the query to service_name = '<name>' and switches the search mode to Traces, so you immediately see the matching traces.

Review service health

  1. Select Traces in the left navigation panel.
  2. Choose Service Catalog from the search-bar toggle.
  3. Select the trace stream and time range you want to inspect.
  4. Review the Status and Error Rate columns to identify services that are degraded, warning, or critical.

Find a specific service

  1. Enter the service name in the filter field.
  2. Review the filtered rows and the match counter above the table.

Investigate a problem service

  1. Identify a service with a warning or critical status, a high error rate, or a P99 highlighted in orange.
  2. Click the service row to open the detail side panel and review its operations.
  3. Select View Traces to open Traces filtered by service_name for that service.
  4. Inspect the matching traces to find the root cause.

Next steps

  • Service Graph: Visualize the same service health information as an interactive topology.
  • Traces: Drill into individual traces and spans for detailed debugging.

Need some help?