Trying to kick off an Architect flow from a Python backend service using the POST /api/v2/flows/executions endpoint. The goal is to have the OpenTelemetry span context travel all the way into the flow so I can see the full trace in Jaeger, including the Data Action calls inside the flow.
Here’s the request payload I’m sending:
import requests
import json
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {access_token}",
# Attempting to inject trace context via headers
"traceparent": "00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01",
"tracestate": "congo=t61rcWkgMzE"
}
body = {
"flowId": "3a4b5c6d-7e8f-9g0h-1i2j-3k4l5m6n7o8p",
"queueId": "9q8r7s6t-5u4v-3w2x-1y0z-a1b2c3d4e5f6",
"participant": {
"id": "some-contact-id",
"externalContactId": "ext-123"
}
}
response = requests.post(url, headers=headers, json=body)
The flow starts fine. I get a 201 Created back with the execution ID. But when I look at the traces for the Data Actions inside that flow, they show up as root spans. The trace IDs don’t match the parent span from my Python app. It looks like Genesys is stripping or ignoring the W3C trace context headers during the execution launch.
Is there a specific way to pass the context? Or does the SDK need to wrap the request differently? I’ve checked the docs and they mention traceId in the body but nothing about header propagation for this specific endpoint.