I'm attempting to implement distributed tracing via OpenCensus across different services where each service is using Flask for serving and sends downstream requests using... er.. requests to build its reply. The tracing platform is GCP Cloud Trace.
I'm using FlaskMiddleware, which traces the initial call correctly, but there's no propagation of span info between the source and target service, even when the middleware has a propagation defined (I've tried a few):
#middleware = FlaskMiddleware(app, exporter=exporter, sampler=sampler, excludelist_paths=['healthz', 'metrics'], propagator=google_cloud_format.GoogleCloudFormatPropagator())
middleware = FlaskMiddleware(app, exporter=exporter, sampler=sampler, excludelist_paths=['healthz', 'metrics'], propagator=b3_format.B3FormatPropagator())
#middleware = FlaskMiddleware(app, exporter=exporter, sampler=sampler, excludelist_paths=['healthz', 'metrics'], propagator=trace_context_http_header_format.TraceContextPropagator())
I guess the question is, does anyone have an example of setting up tracing that traverses several downstream calls when each service is serving via Flask and making downstream requests via requests.
Currently I end up with each service having its own trace with a single span.
