I have a pipeline of asynchronous services communicating by messages which I'm tracing using opencensus.
I have been able to link all the traces from all services by setting the traceID in the span context on each service from metadata that is carried along the pipeline when processing a request.
I also can calculate the total processing time for the whole pipeline in the last step and I know I can add this as an attribute of the span the last step.
What I don't know how to do is to query for traces that belong to a request with a total processing time above a threshold.
What I would like to do is to have a span that starts in the first step of the pipeline and been able to set its duration or latency at the last step. Then I would be able to query for spans with a total latency larger than a threshold.
Maybe I'm missing something, but it seems I'm hitting the same limitation mentioned here Open for event based tracing?
[In open telemetry] there is no standardized way to distribute a span over processes: a span context can be used to create a child span but not for re-creating and continuing the one from which it was retrieved.
