traces

Search slow traces by filtering on duration with a query like @duration:>2000000000, since duration is measured in nanoseconds. Combine this with a service filter and time range, for example service:api @duration:>1000000000 over the last hour.

Filter traces by error type using the @error.type attribute combined with a service filter, such as service:api @error.type:*. This returns all traces containing errors for that service within the selected time window.

Datadog trace search supports service and resource filters, span attributes like @http.status_code, duration filters in nanoseconds, tag search such as env:production, boolean operators AND/OR/NOT, and wildcards like service:web-*.

You need three environment variables: DD_API_KEY for the Datadog API key, DD_APP_KEY for the application key, and DD_SITE for the Datadog site, which defaults to datadoghq.com. Missing credentials produce an error prompting you to export these variables.

Empty results usually mean APM instrumentation is not set up for the service, the query is too restrictive, or the time range is too narrow. Try broadening the query, extending the time window, or verifying that tracing is enabled on the target service.

No, span aggregation and alerting are not supported. For trace analytics monitors use the monitors agent, and for detailed flame graphs or service maps use the Datadog APM UI directly.