How many traces are captured?

The number of traces depends on the sampling rate configured on the Cosmo Router via OTEL instrumentation. Higher sampling rates capture more requests.

What grouping options does Trace Analytics support?

You can group by operation name, client, or error message. The default view shows ungrouped individual requests. Clicking a grouped row applies that group as a filter and returns to the ungrouped view.

What auto-refresh intervals are available?

Auto-refresh intervals are 10 seconds, 30 seconds, 1 minute, and 5 minutes. Auto-refresh is useful during active incidents when you need the data to stay current without manual refreshing.

What filters can I apply?

Available filters depend on the selected date range and grouping. You can filter by operation name, client name and version, and other request attributes.

Can I navigate from a trace to its distributed trace?

Yes. Individual request entries link to the distributed trace view, where you can see the full span tree across the router and all subgraphs.

Does client grouping require any client setup?

Yes. Client grouping reads from the GraphQL-Client-Name and GraphQL-Client-Version request headers (or the Apollo-compatible equivalents). Requests without these headers are still recorded but are not attributed to a named client.

Trace Analytics for Federated GraphQL | Cosmo by WunderGraph

The problem

Request debugging wastes time when data is scattered

Logs across multiple services, no operation context, static snapshots. Debugging a single request should not take an hour.

Log searches take too long during incidents

Finding the requests behind an error spike means searching through distributed logs across multiple services, each with its own format. By the time you find the relevant entries, the incident window has grown.

Log views go stale immediately

Static log snapshots are useless during active incidents. You need the request list to update automatically as new data arrives — without manual refreshing.

Raw request lists are too noisy to read

Thousands of ungrouped requests make it impossible to spot patterns. Without grouping by operation, client, or error, each entry is just noise.

Our solution

A centralized request list with full context

Trace Analytics collects all requests from the Cosmo Router and presents them in a single interface with filtering, grouping, and drill-down. No log searching. No manual correlation across services. Auto-refresh keeps the view current during active incidents.

From request list to root cause

Trace Analytics lists all requests made to your router in a paginated interface.
The amount of traces depends on the sampling rate configured on the Cosmo Router.
Date range selection narrows the list to any period, using predefined ranges or a custom date and time picker.
Filters narrow down to specific requests by operation, client, and other attributes.
Grouping organizes requests by operation name, client, or error message — revealing patterns across large request sets.
Clicking a grouped row applies that group as a filter and returns to the ungrouped individual request view.
Auto-refresh at 10s, 30s, 1min, or 5min keeps data current during active incident monitoring.

From alert to root cause without leaving Cosmo Studio.

Trace Analytics

Before & After

Before Cosmo	With Cosmo
Searching through distributed logs	Centralized list of all requests to the router
Manual correlation of request data	Filters and grouping reveal patterns immediately
Static log views during incidents	Auto-refresh keeps data current
Limited context for each request	Full details including operation, client, and errors

Grouping options

Four ways to view requests

None (default): Each row is an individual request. Full details per entry.
Operation name: Cluster requests by the operation that was executed.
Client: Group by client name and version from request headers.
Error message: Cluster all requests that resulted in the same error. Click through to see individual requests behind each error.

How Trace Analytics works

01

Date, operation, client.

Filter

Narrow the list by date range, operation name, client, or other attributes. Filters can be combined. Applied filters persist as you navigate to other views.

02

Operation, client, or error.

Group

Group by operation name to see which operations appear most. Group by client to analyze per-application behavior. Group by error message to cluster common failures. Clicking any grouped row drills into the individual requests behind it.

03

Full context per request.

Investigate

Click any individual request to see full details: operation performed, client name and version, error message, and timing. Navigate directly to the distributed trace for deeper investigation.

04

10s, 30s, 1min, 5min.

Monitor live

Configure auto-refresh at 10 seconds, 30 seconds, 1 minute, or 5 minutes. During an active incident, the request list updates automatically — useful for confirming whether a deployed fix is working.

What is included

Built into every Cosmo plan

Available on Free, Pro, and Enterprise.

Complete request visibility

Every request made to your federated graph router is listed with operation details, client info, and error messages. Volume is controlled by the router's OTEL sampling rate.

Flexible grouping

Group by operation name, client, or error message. Click any grouped row to apply it as a filter and drill into the individual requests behind it.

Auto-refresh

Configure automatic refresh at 10 seconds, 30 seconds, 1 minute, or 5 minutes. During active incidents, the list updates without manual intervention.

Trace drill-down

Navigate from any request entry to its distributed trace, showing the full span tree across the router and every subgraph it contacted.

Inspect every request to your federated graph

Deploy the Cosmo Router and request traces appear in Cosmo Studio immediately.

Start Free Read the Docs

Debug individual requests with complete trace visibility