Trace-timeline not shown
# questionss
singhe.silva
10/15/2025, 2:22 AMHey!
I am facing some issues when trying to view the trace timeline of promptfoo GUI. I get the following error in the GUI.
“Traces were created but no spans were received. Make sure your provider is:
- Configured to send traces to the OTLP endpoint (http://localhost:4318)
-Creating spans within the trace context
-Properly exporting spans before the evaluation completes.
I am trying to calculate the trace-error-spans of my LLM. But since it didn’t work, I tried writing a FAST API app and view the trace timeline in the GUI.
Can someone help me in this matter please?
m
michael.michael.michael
10/16/2025, 1:10 AMHey! @singhe.silva I looked into this. The tracing system is working correctly - you just need to instrument your provider to send spans to the OTLP endpoint.
When you see "Traces were created but no spans were received", it means Promptfoo created the trace and started the OTLP receiver, but your provider didn't send any spans back.
Here's what you need to do:
1. Install OpenTelemetry packages:
npm install @opentelemetry/api @opentelemetry/sdk-trace-node @opentelemetry/exporter-trace-otlp-http @opentelemetry/sdk-trace-base
2. In your provider, initialize OpenTelemetry at the module level:
const { trace, context, SpanStatusCode } = require('@opentelemetry/api');
const { NodeTracerProvider } = require('@opentelemetry/sdk-trace-node');
const { OTLPTraceExporter } = require('@opentelemetry/exporter-trace-otlp-http');
const { BatchSpanProcessor } = require('@opentelemetry/sdk-trace-base');
const provider = new NodeTracerProvider();
const exporter = new OTLPTraceExporter({
url: 'http://localhost:4318/v1/traces',
});
const spanProcessor = new BatchSpanProcessor(exporter);
provider.addSpanProcessor(spanProcessor);
provider.register();
const tracer = trace.getTracer('my-provider');
michael.michael.michael
10/16/2025, 1:11 AM4. Make sure your config has tracing enabled and your tests have the metadata:
tracing:
enabled: true
otlp:
http:
enabled: true
port: 4318
tests:
- vars:
topic: 'example'
metadata:
tracingEnabled: true
The most common mistake is forgetting the forceFlush() call. Without it, spans don't export before the evaluation completes.
For Python/FastAPI:
from opentelemetry import trace
from opentelemetry.propagate import extract
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
provider = TracerProvider()
exporter = OTLPSpanExporter(endpoint="http://localhost:4318/v1/traces")
provider.add_span_processor(BatchSpanProcessor(exporter))
trace.set_tracer_provider(provider)
tracer = trace.get_tracer(__name__)
def call_api(prompt, context):
if 'traceparent' not in context:
return {"output": your_llm_call(prompt)}
ctx = extract({"traceparent": context["traceparent"]})
with tracer.start_as_current_span("provider.call", context=ctx) as span:
result = your_llm_call(prompt)
provider.force_flush()
return {"output": result}
Check out the working example in the repo: examples/opentelemetry-tracing/
Let me know if you run into any issues.
michael.michael.michael
10/16/2025, 1:11 AM3. In your callApi function, parse the traceparent and create spans:
async callApi(prompt, promptfooContext) {
if (!promptfooContext?.traceparent) {
return { output: await yourLLMCall(prompt) };
}
// Parse traceparent
const matches = promptfooContext.traceparent.match(
/^(\d{2})-([a-f0-9]{32})-([a-f0-9]{16})-(\d{2})$/
);
if (!matches) return { output: await yourLLMCall(prompt) };
const [, , traceId, parentId, traceFlags] = matches;
// Create parent context
const parentCtx = trace.setSpanContext(context.active(), {
traceId,
spanId: parentId,
traceFlags: parseInt(traceFlags, 16),
isRemote: true,
});
// Run within trace context
return context.with(parentCtx, async () => {
const span = tracer.startSpan('provider.call');
try {
const result = await yourLLMCall(prompt);
span.setStatus({ code: SpanStatusCode.OK });
// CRITICAL: Force flush before returning
await spanProcessor.forceFlush();
return { output: result };
} catch (error) {
span.recordException(error);
span.setStatus({ code: SpanStatusCode.ERROR, message: error.message });
throw error;
} finally {
span.end();
}
});
}
s
singhe.silva
10/16/2025, 4:20 AMThank you Michael, will try this and get back to you!
2 Views